assessing how delayed entry program physical fitness is
Post on 23-Nov-2021
6 Views
Preview:
TRANSCRIPT
Distribution limited to Sponsor.
Assessing How Delayed Entry
Program Physical Fitness is
Related to In-Service Attrition,
Injuries, and Physical Fitness
September 2014
(b) (6)
(b) (6)
(b) (6)
(b) (6)
(b) (6)
(b) (6)
(b) (6)
(b) (6)(b) (6)
(b) (6)
Copyright © 2014 CNA
This document contains the best opinion of CNA at the time of issue.
It does not necessarily represent the opinion of the sponsor.
Distribution
Distribution limited to Sponsor. Specific authority: N00014-11-D-0323.
Copies of this document can be obtained through the Defense Technical Information
Center at www.dtic.mil or contact CNA Document Control and Distribution Section
at 703-824-2123.
Photography Credit: Marines demonstrate how to properly perform crunches for an
Initial Strength Test for female poolees who are enrolled in the Delayed Entry Program.
Westover Air Reserve Base, Chicopee, Massachusetts, Mar. 25, 2013. (U. S. Marine Corps
photo by /Released)
Approved by: September 2014
Marine Corps Manpower Team
Resource Analysis Division
(b) (6)
(b) (6)
i
Abstract
Over the past several years, the Department of Defense has asked
the services to investigate their ability to expand opportunities for
women in the military. In support of this initiative, the Marine
Corps started a deliberate and measured effort to examine the
possible integration of female Marines into ground combat units
and Military Occupational Specialties (MOSs) with the development
of the Marine Corps Force Integration Plan (MCFIP). In support of
this effort, CNA was asked to examine the relationship between the
Initial Strength Test (IST) given to recruits at the time of enlistment
and early attrition, recruit training injury rates, scores on the
Physical Fitness Test (PFT), and scores on the Combat Fitness Test
(CFT); and how these relationships vary by gender. This paper
presents the results of this examination. We found that the IST
score is a good predictor of attrition, injury rates, and PFT and CFT
scores, with a higher IST score leading both to lower attrition and
injury rates and to higher PFT and CFT scores. We also found,
however, that a significant share of men and women who score well
on the IST end up scoring poorly on the PFT and CFT; conversely, a
significant share who score poorly on the IST, score well on the PFT
and CFT. This latter finding suggests that any classification
policies for physically-demanding MOSs that are based on IST
scores should include provisions to reconsider the MOS
classification if recruit training PFT and CFT scores differ
significantly from the IST score.
ii
This page intentionally left blank.
iii
Executive Summary
The recent lifting of Department of Defense (DOD) combat exclusion provisions has
prompted the Marine Corps to consider administering physical fitness tests before
Marines are classified into physically demanding Military Occupational Specialties
(MOSs). Since this classification process starts in the Delayed Entry Program (DEP),
the Marine Corps needs to know whether the Initial Strength Test (IST) administered
in the DEP can be used to classify recruits into Programs Enlisted For (PEFs) that
contain physically demanding MOSs. Central to answering this question is whether
IST scores are reliable predictors of attrition, injuries during recruit training, and in-
service physical fitness. To assist the Marine Corps with early classification of
recruits into physically demanding MOSs, this paper explores the relationships
between DEP IST scores and attrition, injuries during recruit training, and in-service
physical fitness.
Since the Marine Corps is considering using the IST as a criterion for PEF assignment,
we selected the first IST score recorded in the DEP, since the timing would generally
coincide with assignment of a recruit’s PEF. Recruiters try to assign PEFs to recruits
as early as possible once they are in the DEP, so we chose the first IST recorded in the
DEP to facilitate our analysis. For the DEP IST score, we combined the three separate
IST events—the 1.5-mile run, pull-ups/flexed-arm hang (FAH), and crunches—to
create a composite score that, as we explain later in this report, closely mirrors the
scoring procedure for the Physical Fitness Test (PFT). We then measured the
relationship between this composite DEP IST score and attrition rates, injury rates
while in recruit training, and various PFT and Combat Fitness Test (CFT) scores taken
during a Marine’s initial enlistment.
DEP IST score and attrition rates
Table 1 summarizes the relationship that we found between the DEP IST score and
attrition during three key time periods in an enlisted Marine’s career:
during recruit training,
before the completion of 24 months of service, given that the Marine
successfully completes recruit training, and
iv
before 45 months of service, given that the Marine successfully completes 24
months of service.
From Table 1, we see that, although the average attrition rate for men during recruit
training is 7.6 percent, the predicted recruit training attrition rate for men who score
in the top third of all male recruits on the IST is only 5.0 percent. This rate rises to
7.1 percent for men who score in the middle third of the IST, and to 10.2 percent for
men who score in the bottom third of the IST. For women, the relationships are
similar: higher scores on the DEP IST predict lower attrition rates during recruit
training for women.
Table 1. Predicteda recruit training attrition rates, conditional 24-month attrition
rates, and conditional 45-month attrition rates by DEP IST categories and
gender
Men Women
Recruit
training
attrition
Attrition before 24
months of service,
conditional on
completing recruit
training
Attrition before 45
months of service,
conditional on
completing 24
months of service
Recruit
Training
Attrition
Attrition before 24
months of service,
conditional on
completing
recruit training
Attrition before 45
months of service,
conditional on
completing 24
months of service
IST top third 5.0% 6.1% 5.8% 10.1% 7.4% 7.2%
IST middle third 7.1% 6.8% 6.6% 14.1% 8.7% 7.6%
IST bottom
third
10.2% 7.1% 7.8% 19.6% 9.8% 8.0%
Average
attrition
(actual)
7.6% 6.7% 6.8% 14.7% 8.7% 7.6%
Number of
observationsb
257,385 190,026 134,880 21,910 14,530 9,676
Data source: CNA Marine Corps personnel files
a The predictions are based on coefficient estimates in a logit regression. These coefficient
estimates are given in Table 14, Table 17, and Table 18 in the appendix.
b. The number of observations progressively drops from recruit training to 24-month to 45-
month attrition. This is because some recruits who could be counted for recruit training
attrition cannot be counted for 24- and 45-month attrition because they will not have
been on active duty long enough to have reached these benchmarks.
The 24- and 45-month conditional attrition rates show whether the effect of the IST
score on attrition is persistent throughout the course of a Marine’s initial enlistment
or whether that effect declines over time. For both men and women, higher IST
scores are associated with decreased attrition over the course of a Marine’s first
enlistment, but the relative effect of these higher IST scores on attrition rates
generally declines throughout the first enlistment.
v
DEP IST score and injury rates
We were unable to obtain injury data, so we proxy recruit training injuries as follows:
injured recruits are defined as those who were discharged for medical reasons or
those who were medically recycled during recruit training.1 To compare men and
women, we analyze only recruits trained at Parris Island, since women are only
trained at Parris Island.
We find that injury rates during recruit training are lower for both men and women
who have higher DEP IST scores. Although the average injury rate for men in recruit
training is 3.6 percent, men in the top third of the IST score have only a 2.5 percent
predicted injury rate, men in the middle third have a 3.7 percent predicted injury
rate, and men in the bottom third have a 4.7 percent predicted injury rate in recruit
training. The average female injury rate is 6.0 percent, and the corresponding
percentages follow a similar pattern: 4.7, 5.9, and 7.7. Thus, we conclude that higher
DEP IST scores predict lower injury rates in recruit training for both men and women.
DEP IST score and PFT and CFT scores
To measure the relationship between the DEP IST score and various PFT scores taken
during a Marine’s first enlistment, we classify men and women separately into one of
the following categories:
IST 300 (perfect score),
IST top third (but not 300),
IST middle third, and
IST bottom third.
From Table 2, we see that although the average score for the PFT taken at the end of
recruit training for men is 244, the predicted PFT score for men scoring 300 on the
DEP IST is 291, the predicted PFT score for men scoring in the top third (but not 300)
1 The expected timeline to receive actual injury data was too long for this quick-response study.
We consider the injury indicator that we constructed a sufficient proxy for actual injury data.
vi
is 264, the predicted PFT score for men scoring in the middle third is 244, and the
predicted PFT score for men scoring in the bottom third is only 224.
For women, the relationships are similar to those for men. Although the average
score for the PFT taken at the end of recruit training for women is 250, the predicted
PFT score for women scoring 300 on the DEP IST is 291, the predicted PFT score for
women scoring in the top third (but not 300) is 267, the predicted PFT score for
women scoring in the middle third is 251, and the predicted PFT score for women
scoring in the bottom third is only 232.
We conclude, on average, that, for both women and men, recruits who have high
scores on the DEP IST are predicted to have high scores on the PFT taken at the end
of recruit training as well.
Table 2. Predicteda recruit training, first year, and second PFT scores by IST
categories and gender
Men Women
Recruit
training
PFT
First year
PFT
Second year
PFT
Recruit
training
PFT
First year
PFT
Second year
PFT
300 IST 291 289 289 291 286 285
IST top third
(but not 300)
264 260 264 267 255 259
IST middle
third
244 240 245 251 234 240
IST bottom
third
224 220 225 232 215 222
Average 244 240 244 250 235 241
Number of
observations
229,635 179,287 171,055 16,824 13,034 12,357
Data Source: CNA Marine Corps personnel files a The predictions are based on coefficient estimates in a tobit regression. These coefficient
estimates are given in Table 20 to Table 22 in the appendix.
Although the DEP IST is a good predictor of PFT scores in the first two years of a
Marine’s initial enlistment, a significant number of men and women go from a low
DEP IST to a high PFT, and from a high DEP IST to a low PFT. For example, we find
that:
13.1 percent of the women who score in the bottom third of the DEP IST later
score in the top third (but not 300) PFT at the end of recruit training.
13.5 percent of the women who score in the top third (but not 300) of the DEP
IST later score in the bottom third of the PFT at the end of recruit training.
vii
The findings for men were similar. These changes are due to both relative and
absolute changes in Marines’ measured fitness levels.
This suggests that MOS classification may need to be adjusted once final fitness
scores are available at the end of recruit training. Otherwise, some men and women
whose physical fitness level is lower than the DEP IST indicated may not succeed in
physically demanding MOSs—and some who would succeed in those MOSs will miss
opportunities because their DEP IST scores do not indicate their in-service physical
fitness level.
We also examined the relationship between the DEP IST score and scores for various
CFTs taken during a Marine’s first enlistment, and found similar results.
viii
This page intentionally left blank.
ix
Contents
Introduction ............................................................................................................................ 1
Data .......................................................................................................................................... 2
Methodology ........................................................................................................................... 2
Relationship between DEP IST score and three attrition rates ............................. 2
Relationship between DEP IST score and recruit training injury rates ............... 4
Relationship between DEP IST score and PFT and CFT scores ............................. 4
Relationship Between DEP IST Scores and Attrition ....................................................... 6
Relationship between DEP IST score and recruit training attrition ............................ 6
Relationship between DEP IST score and 24-month attrition ...................................... 8
Relationship between DEP IST score and 45-month attrition .................................... 10
Relationship between DEP IST score and conditional 24-month and 45-month attrition ................................................................................................................................. 11
Relationship Between DEP IST Scores and Recruit Training Injuries ........................ 14
Relationship Between DEP IST Scores and PFT and CFT Scores ................................ 16
Relationship between DEP IST score and final recruit training PFT score .............. 16
Relationship between DEP IST score and first annual PFT score .............................. 18
Relationship between DEP IST score and second annual PFT score ......................... 21
Relationship between DEP IST score and final recruit training CFT score.............. 23
Relationship between DEP IST score and first annual CFT score ............................. 25
Relationship between DEP IST score and second annual CFT score ........................ 28
Conclusions .......................................................................................................................... 31
Appendix: Logit and Tobit Regression Results .............................................................. 33
x
This page intentionally left blank.
xi
List of Tables
Table 1. Predicteda recruit training attrition rates, conditional 24-month
attrition rates, and conditional 45-month attrition rates by DEP IST
categories and gender .................................................................................... iv Table 2. Predicteda recruit training, first year, and second PFT scores by IST
categories and gender .................................................................................... vi Table 3. Recruit training attrition rates by DEP IST categories and gender,
FY05–FY13 accessions .................................................................................... 8 Table 4. 24-month attrition rates by DEP IST categories and gender, FY05 –
FY12 accessions ............................................................................................... 9 Table 5. 45-month attrition rates by DEP IST categories and gender, FY05 –
FY10 accessions ............................................................................................. 11 Table 6. Conditional attrition probabilities to 24 and 45 months by IST
category and gender ...................................................................................... 12 Table 7. Recruit training injury rates by DEP IST categories and gender,
FY05 –FY13 accessions ................................................................................. 14 Table 8. Relationship between DEP IST categories and recruit training PFT
categories by gender, FY05-FY13 accessions ........................................... 17 Table 9. Relationship between DEP IST categories and first year PFT
categories by gender, FY05-FY12 accessions ........................................... 20 Table 10. Relationship between DEP IST categories and second year PFT
categories by gender, FY05-FY11 accessions ........................................... 22 Table 11. Relationship between DEP IST categories and recruit training CFT
categories by gender, FY05-FY13 accessions ........................................... 24 Table 12. Relationship between DEP IST categories and first year CFT
categories by gender, FY05-FY12 accessions ........................................... 27 Table 13. Relationship between DEP IST categories and second year CFT
categories by gender, FY05-FY11 accessions ........................................... 29 Table 14. Recruit training attrition logit regression estimates by gender,
FY05-FY13 accessions ................................................................................... 33 Table 15. 24-month attrition logit regression estimates by gender, FY05-
FY12 accessions ............................................................................................. 35 Table 16. 45-month attrition logit regression estimates by gender, FY05-
FY10 accessions ............................................................................................. 37 Table 17. 24-month conditional attrition logit regression estimates by
gender, FY05-FY12 accessions .................................................................... 39
xii
Table 18. 45-month conditional attrition logit regression estimates by
gender, FY05-FY10 accessions .................................................................... 41 Table 19. Recruit training injury logit regression estimates by gender, FY05-
FY13 accessions ............................................................................................. 43 Table 20. Recruit training PFT tobit regression estimates by gender, FY05-
FY13 accessions ............................................................................................. 45 Table 21. First year PFT tobit regression estimates by gender, FY05-FY12
accessions ........................................................................................................ 47 Table 22. Second year PFT tobit regression estimates by gender, FY05-FY11
accessions ........................................................................................................ 49 Table 23. Recruit training CFT tobit regression estimates by gender, FY05-
FY13 accessions ............................................................................................. 51 Table 24. First year CFT tobit regression estimates by gender, FY05-FY12
accessions ........................................................................................................ 53 Table 25. Second year CFT tobit regression estimates by gender, FY05-FY11
accessions ........................................................................................................ 55 Table 26. DEP attrition logit regression estimates by gender, FY05-FY13
contracts .......................................................................................................... 57
xiii
Glossary
CFT Combat Fitness Test
DEP Delayed Entry Program
DOD Department of Defense
FAH Flexed-Arm Hang
IST Initial Strength Test
OLS Ordinary Least Squares
MCFIP Marine Corps Force Integration Plan
MCRD Marine Corps Recruit Depot
MCRISS Marine Corps Recruiting Information Support System
MCTIMS Marine Corps Training Information Management System
MOS Military Occupational Specialty
PEF Program Enlisted For
PFT Physical Fitness Test
TFDW Total Force Data Warehouse
xiv
This page intentionally left blank.
1
Introduction
Now that the Department of Defense (DOD) has lifted combat exclusion provisions
for women, the Marine Corps has been conducting a deliberate and measured effort
to integrate women into ground combat units and Military Occupational Specialties
(MOSs). Part of this effort has focused on understanding the physical requirements
for closed MOSs and units, and how those requirements should inform changes to
the Marine Corps’ personnel classification and assignment policies. A key question in
this effort has been: What physical fitness measures most reliably proxy the physical
demands of duties in closed MOSs and units? Measurement options being considered
have ranged from proxy tests uniquely tailored to the physical tasks for closed MOSs,
to some combination of the individual events that are components of various Marine
Corps physical fitness tests, including the Initial Strength Test (IST), Physical Fitness
Test (PFT), and Combat Fitness Test (CFT). As discussed in this report, CNA’s
research is to exploring these options.
Here, we summarize CNA’s research on the relationship between the first IST that an
enlisted recruit takes while in the Delayed Entry Program (DEP) and early attrition;
the relationship between the IST and injury rates in recruit training; and the
relationship between the IST and the PFT and the CFT taken in recruit training and in
the first and second years of service. These relationships are estimated separately for
men and women. This report is the deliverable for Task 3A of the study CNA Support
to the Marine Corps Force Integration Plan (MCFIP). The analysis is related to
potential policies that could include the Marine Corps using a version of the DEP IST
score as a criterion for classification into Programs Enlisted For (PEF) that contain
physically demanding MOSs.2
2 The IST was established in the early 1970s to provide a standard for determining whether
recruits were physically ready to start recruit training. To date, it has not been used as a
criterion for classification decisions or for any other recruit-related decisions.
2
Data
The data for DEP IST scores and attrition in the DEP are extracted from the Marine
Corps Recruiting Information Support System (MCRISS). The data on attrition once
the Marine enters the Marine Corps Recruit Depot (MCRD), and the data on all the
PFTs and CFTs a Marine takes while in the Marine Corps, are extracted from Marine
Corps Total Force Data Warehouse (TFDW) snapshots. We construct a proxy variable
for whether a recruit is injured in recruit training, using a combination of MCRISS
and Marine Corps Training Information Management System (MCTIMS) data. A
recruit is defined as injured if he or she was discharged for medical reasons from
recruit training or was medically recycled in his or her first attempt at recruit
training.
Methodology
To study the relationship between the DEP IST and different indicators of a Marine’s
success in his or her first enlistment, we combine the three separate IST events—the
1.5-mile run, pull-ups/flexed-arm hang (FAH), and crunches—of the first IST test
taken in DEP, to construct a composite score.3 We measure the relationship between
the gender-normed IST score and attrition rates, injury rates while in recruit training,
and various gender-normed PFT and CFT scores taken during a Marine’s initial
enlistment.
Relationship between DEP IST score and three attrition
rates
The variable of interest when analyzing attrition is a zero or one indicator variable
that takes on a value of one if an enlisted Marine attrites during a specified time
period in his or her career, and takes on a value of zero if the recruit successfully
3 The maximum composite score for the IST is 300, the same as the maximum score for the
PFT. For the crunches and pull-up/FAH components, the score for the IST is computed exactly
the same as for the PFT, with a maximum of 100 points in each component, but, unlike the
computation of PFT scores, there is no lower bound of 3 pull-ups (15 points), 40 crunches (40
points), and 15 seconds (15 points) on the FAH. For the 1.5-mile run, a time of 9 minutes or
less for men and 10.5 minutes or less for women is given 100 points, with one point deducted
for each additional five seconds over the respective 9-minute and 10.5-minute thresholds.
3
completes that time period in his or her career. We define three attrition indicator
variables to capture three key time periods4 in an enlisted Marine’s career:
during recruit training,
before the completion of the first 24 months of service, and
before the completion of the first 45 months of service.
To measure the relationship between the DEP IST and attrition during recruit
training, we use two approaches. (The relationships between the DEP IST score and
24- and 45-month attrition are measured similarly.)
In the first approach, we classify5 each recruit as being in the top third, the middle
third, or the bottom third of the DEP IST scores of all recruits who accessed. This
classification is done separately for men and women.
We then perform cross tabulations between the respective three thirds of IST scores
and the attrition indicator variable for recruit training. These cross tabulations yield
the following three percentages:
1. the percentage of Marines who were in the top third of DEP IST scores and
attrited during recruit training;
2. the percentage of Marines who were in the middle third of DEP IST scores and
attrited during recruit training; and
3. the percentage of Marines who were in the bottom third of DEP IST scores and
attrited during recruit training.
In the second approach, we run a logit6 regression, where the observed recruit
training attrition indicator variable is the dependent variable and the DEP IST score is
4 We also examined the relationship between IST score and DEP attrition, but 48 percent of DEP
attrites were missing an IST score and we did not feel that the available data on IST scores for
DEP attrites allowed for an in-depth analysis of the effect of IST score on DEP attrition. We did,
however, analyze the relationship between DEP attrition and IST score using the data on the 52
percent of DEP attrites that did have an IST score. These results are presented in the appendix,
in Table 26.
5 For men and women who enter recruit training, a recruit is classified as being in the bottom
third of the IST if his IST score was 163 or less (IST ≤ 163); if his IST score was higher than 163
but lower than 208 (164 ≤ IST ≤ 207) he was classified as being in the middle third; and if his
IST was greater than 207 (IST ≥ 208) he was classified as being in the top third of the IST. For
women, these respective cut-off points are: IST ≤ 154 for the bottom third; 155 ≤ IST ≤ 200 for
the middle third; and IST ≥ 201 for the top third.
4
the main independent variable of interest. We also include variables in these logit
regressions to control for other characteristics and factors that have been shown to
affect attrition. These include: whether the recruit is high quality (i.e., a high school
diploma graduate with a score of 50 or higher on the Armed Forced Qualification
Test); the recruit’s age, race/ethnicity, enlistment waivers, enlistment bonuses, and
time in the DEP; the time of year at which the recruit was accessed; and the MCRD
(i.e., San Diego or Parris Island) where the recruit was trained.7
A similar set of cross tabulations and logit regressions are done to measure the
relationship between DEP IST score and 24- and 45-month attrition.
Relationship between DEP IST score and recruit
training injury rates
For the injury indicator variable, we perform cross tabulations and conduct logit
regression analysis similar to those described above for attrition. This indicator
variable takes on a value of one if the recruit was injured during recruit training, and
zero if the recruit was not injured during recruit training.
Relationship between DEP IST score and PFT and CFT
scores
We first describe our methodology for measuring the relationship between DEP IST
scores and PFT scores attained at the end of recruit training. The relationship
between DEP IST scores and CFT scores attained at the end of recruit training, as well
as the relationships between DEP IST scores and the first and second year PFT and
CFT scores, are measured similarly. We measure all of these relationships with two
approaches.
In the first approach, we classify each recruit as being in the top third, the middle
third, or the bottom third of DEP IST scores of all recruits. In the top third, we
further classify recruits who achieved the maximum score of 300 on the DEP IST. If a
recruit’s DEP IST score falls in the top third of all scores (but not 300), the recruit is
classified as an IST top third recruit, and so forth. Finally, if the recruit achieves the
maximum score of 300 on the DEP IST, the recruit is classified as an IST 300 recruit.
6 The logit regression is similar to an ordinary least squares (OLS) regression, except that the
logit regression has a non-linear rather than a linear functional form and the logit regression
accounts for the fact that the observed attrition variable is bounded by zero and one.
7 These same control variables are used for all logit and tobit regressions.
5
This gives us four categories8 of recruits: (1) IST 300, (2) IST top third, (3) IST middle
third, and (4) IST bottom third. This classification is done separately for men and
women.
We perform this same classification of each Marine as being in either the top third
(but not 300), middle third, or bottom third of observed PFT scores for all recruits at
the end of recruit training. There also is a fourth category for recruits who have a
PFT score of 300.9 This classification is done separately for men and women. We then
perform cross tabulations between the respective four categories of DEP IST scores
and the four categories of recruit training PFT scores.
The second approach is to run a tobit10 regression, where the observed PFT
composite score is the dependent variable and the DEP IST score is the main
independent variable of interest.
A similar set of cross tabulations are constructed and tobit regressions estimated for
the CFT scores taken at the end of recruit training, as well as for the PFT and CFT
scores taken during the first and second years of a Marine’s initial enlistment.
8 For men who take the final PFT in recruit training, a recruit is classified as being in the
bottom third of the IST if his IST score was 165 or less (IST ≤ 165); if his IST score was higher
than 165 but lower than 210 (166 ≤ IST ≤ 209) he is classified as being in the middle third; if
his IST was greater than 209 but less than 300 (210 ≤ IST ≤ 299) he is classified as being in the
top third of the IST, and if he scored a 300 he is classified as an IST 300 recruit. For women,
these respective cut-off points are: IST ≤ 158 for the bottom third; 159 ≤ IST ≤ 204 for the
middle third; 205 ≤ IST ≤ 299 for the top third; and IST 300.
9 For men, a recruit is classified as being in the bottom third of the final recruit training PFT if
his PFT score was 232 or less (PFT ≤ 232); if his PFT score was higher than 232 but lower than
262 (233 ≤ PFT ≤ 261) he is classified as being in the middle third; if his PFT was greater than
261 but less than 300 (262 ≤ PFT ≤ 299) he is classified as being in the top third of the PFT; and
if he scored a 300 he is classified as being a 300 PFT recruit. For women, these respective cut-
off points are: PFT ≤ 242 for the bottom third; 243 ≤ PFT ≤ 269 for the middle third; and 270 ≤
PFT ≤ 299 for the top third; and a 300 PFT recruit.
10 The tobit regression is similar to an OLS regression, where the tobit regression accounts for
the fact that observed PFT and CFT scores are censored from above by a score of 300 and
censored from below by a score of 134.
6
Relationship Between DEP IST Scores
and Attrition
In this section, we discuss the relationship between the DEP IST score and the
attrition of Marine recruits in three11 key time periods of their careers: (1) during
recruit training, (2) before the completion of the first 24 months of service, and (3)
before the completion of the first 45 months of service.
Relationship between DEP IST score and
recruit training attrition
The relationship between the DEP IST score and attrition during recruit training at
MCRD Parris Island or San Diego is summarized in Table 3 below and in Table 14 in
the appendix.
Table 3 shows the relationship between three categories12 of DEP IST scores (top
third, middle third, and bottom third) and attrition rates in recruit training for both
women and men.
From Table 3, we see that of the 257,398 men who entered recruit training between
FY05 and FY13, 7.6 percent attrited before completing recruit training. Of the 21,923
women who entered recruit training between FY05 and FY13, 14.8 percent attrited
before completing recruit training.
Although overall recruit training attrition for men was 7.6 percent, the attrition rate
was only 4.7 percent for men who scored in the top third of the DEP IST and was 10.9
11 We also analyzed the relationship between DEP attrition and the IST score. These results are
summarized in Table 26 in the appendix.
12 For men who entered recruit training, a recruit is classified as being in the bottom third of
the IST if his IST score was 163 or less (IST ≤ 163); if his IST score was higher than 163 but
lower than 208 (164 ≤ IST ≤ 207) he is classified as being in the middle third; and if his IST was
greater than 207 (IST > 208) he is classified as being in the top third of the IST. For women,
these respective cut-off points are: IST ≤ 154 for the bottom third; 155 ≤ IST ≤ 200 for the
middle third; and IST > 201 for the top third.
7
percent for those who scored in the bottom third of the DEP IST. We find a similar
pattern of attrition for women in Table 3. Although the overall recruit training
attrition for women was 14.8 percent, from Table 3 we see that this attrition rate was
only 9.8 percent for women who scored in the top third of the DEP IST and was 20.2
percent for women who scored in the bottom third.
From Table 3, we can conclude that the IST taken in DEP is a good predictor of
success in recruit training. We investigate this relationship further with a logit
regression equation, where the dependent variable is a zero or one indicator variable
for recruit training attrition. The logit regression allows us to formally test whether
the IST categories13 shown in Table 3 are good predictors of recruit training attrition,
while controlling for other factors that influence recruit training attrition, such as
aptitude test scores, enlistment bonus, and accession year. These logit regression
results are shown in Table 14 of the appendix.
From the coefficient estimates in Table 14 in the appendix, we see that being in the
middle third or bottom third of the DEP IST scores increases attrition relative to the
top IST group for both men and women.14 The marginal changes (labeled “derivative”
in the tables in the Appendix)15 for the two IST category variables shown in Table 14
13 We capture the impact of IST scores on recruit training attrition with two IST score indicator
variables. The first IST score indicator variable, labeled “IST middle third,” takes a value of 1 if
the recruit scored in the middle third of the IST score and a value of zero otherwise. The
second indicator variable to capture the impact of IST score on recruit training attrition is
labeled “IST bottom third.” This indicator variable takes a value of 1 if the recruit scored in the
bottom third of the IST score, and zero otherwise. These two indicator variables allow us to
compare these two categories of IST scores to the top third IST score category, which is in the
“base group” of the logit regression.
14 The z-statistics for both these coefficient estimates indicate that the coefficient estimates are
statistically significantly different from zero. All reported differences from tobit and logit
regression estimates throughout the remainder of this paper are statistically significant unless
otherwise indicated.
15 For men, the “marginal change” shown in Table 14 in the Appendix (labeled “derivative” in
the tables in the Appendix) for the IST middle third and IST bottom third coefficients were
computed as follows. First, based on the coefficients given in Table 14 for men, we predict the
probability of attrition for each man in the sample assuming he was in the top third of the IST
distribution, and then compute the average of these predictions. Doing this for the sample of
257,385 men in our dataset, we get an average predicted attrition rate of 5.1 percent if all men
in the sample were in the top third IST group. Second, based on the coefficients given in Table
14 for men, we predict the probability of attrition for each man in the sample assuming he was
in the middle third of the IST distribution, and then compute the average of these predictions.
Doing this for the sample of 257,385 men in our dataset, we get an average predicted attrition
rate of 7.1 percent if all men in the sample were in the middle third IST score. Third, based on
the coefficients given in Table 14 for men, we predict the probability of attrition for each man
in the sample assuming he was in the bottom third of the IST distribution, and then compute
the average of these predictions. Doing this for the sample of 257,385 men in our dataset, we
get an average predicted attrition rate of 10.2 percent if all men in the sample were in the
8
imply that male recruits in the IST middle third have a 2-percentage-point higher
attrition rate, and those in the IST bottom third have a 5.2-percentage-point higher
attrition rate, than those in the IST top third. The marginal changes of the two IST
category variables in Table 14 for female recruits imply that female recruits in the
IST middle third have a 4.1-percentage-point higher attrition rate, and those in the
IST bottom third have a 9.6-percentage-point higher attrition rate, than those in the
IST top third.
Table 3. Recruit training attrition rates by DEP IST categories and gender,
FY05–FY13 accessions
Men Women
Total
recruits Attrition
Total
recruits Attrition
IST top third 84,440 4.7% 7,244 9.8%
IST middle third 86,395 7.1% 7,319 14.3%
IST bottom third 86,563 10.9% 7,360 20.2%
Total 257,398 7.6% 21,923 14.8%
Source: CNA Marine Corps personnel files
Relationship between DEP IST score and 24-
month attrition
The relationship between the DEP IST score and attrition during the first 24 months
of service in the Marine Corps is summarized in Table 4 below and in Table 15 in the
appendix.
Table 4 shows the relationship between three categories of DEP IST scores (top third,
middle third, and bottom third) and the overall attrition rates in the first 24 months
of service16 for both male and female Marines. As with recruit training attrition, the
bottom third IST group. Finally, taking the difference between 7.1 percent and 5.1 percent, we
get the 2.0 percent impact of being in the middle third versus the top third IST group shown in
Table 14; and taking the difference between 10.2 percent and 5.1 percent, we get the 5.1
percent impact of being in the bottom third of the IST group shown in Table 14.
16 We are looking at total attrition over the 24-month period, not attrition conditional on having
completed recruit training. So while 22.3 percent of the 17,311 women in the sample attrited
over the 24-month period, 16.0 percent of these women attrited in recruit training, and only 7.3
percent of them attrited between recruit training and 24 months of service. For the 207,364
9
relationship between the three categories of IST scores and overall attrition
continues to be strong after 24 months of service. Moreover, the differences between
the three IST categories are as large as, or even larger than, they were for recruit
training attrition. This suggests that the DEP IST captures fairly persistent
differences in attrition over time.
Table 4. 24-month attrition rates by DEP IST categories and gender, FY05 –FY12
accessions
Men Women
Total
recruits Attrition
Total
recruits Attrition
IST top third 63,815 10.0% 5,427 16.2%
IST middle third 69,222 13.2% 5,771 21.9%
IST bottom third 74,327 17.1% 6,113 28.1%
Total 207,364 13.6% 17,311 22.3%
Source: CNA Marine Corps personnel files
From Table 4, we can conclude that the IST taken in DEP is a good predictor of
success in the first two years of service. As we did with recruit training attrition, we
investigate this relationship further with a logit regression equation, where the
dependent variable is a zero or one indicator variable for overall attrition in the first
24 months of service. These logit regression results are shown in Table 15 of the
appendix.
From the coefficient estimates in Table 15 in the appendix, we see that both men and
women in the IST middle third or IST bottom third have higher 24-month attrition
rates than those in the IST top third. The marginal changes for the two IST category
variables shown in Table 15 imply that male recruits in the IST middle third have a
2.4-percentage-point higher overall 24-month attrition rate, and those in the IST
bottom third have a 5.7-percentage-point higher overall 24-month attrition rate, than
those in the IST top third. The corresponding marginal changes for female recruits
imply that female recruits in the IST middle third have a 5.1-percentage-point higher
overall 24-month attrition rate, and those in the IST bottom third have an 11.0-
men in the sample, 13.6 percent of them attrited over the 24-month period, but 8.4 percent of
these men attrited in recruit training, and only 5.2 percent of them attrited between recruit
training and 24 months of service.
10
percentage-point higher overall 24-month attrition rate, than those in the IST top
third.
Relationship between DEP IST score and 45-
month attrition
The relationship between the DEP IST score and attrition during the first 45 months
of service in the Marine Corps is summarized in Table 5 below and in Table 16 of the
appendix.
Table 5 shows the relationship between the three categories of DEP IST scores and
attrition rates in the first 45 months of service17 for both male and female Marines.
As with recruit training attrition and overall 24-month attrition, the relationship
between the three categories of IST scores and attrition continues to be strong after
45 months of service.
17 We are looking at total attrition over the 45-month period, not attrition conditional on having
completed 24 months of service. While 29.9 percent of the 12,938 women in the sample
attrited over the 45-month period, 17.2 percent of these women attrited in recruit training, 7.0
percent of these women attrited between recruit training and 24 months of service, and only
5.7 percent of them attrited between 24 and 45 months of service. For the 159,625 men in the
sample, 20.3 percent of them attrited over the 45-month period, but 9.0 percent of these men
attrited in recruit training, 5.5 percent of these men attrited between recruit training and 24
months of service, and only 5.8 percent of them attrited between 24 and 45 months of service.
For women, 57.5 percent of 45-month attrition occurred during recruit training—for men, this
percentage was only 44.3.
11
Table 5. 45-month attrition rates by DEP IST categories and gender, FY05 –FY10
accessions
Men Women
Total
recruits Attrition
Total
recruits Attrition
IST top third 44,003 15.7% 3,758 23.4%
IST middle third 52,545 19.6% 4,279 29.4%
IST bottom third 63,077 24.2% 4,901 35.2%
Total 159,625 20.3% 12,938 29.9%
Source: CNA Marine Corps personnel files
From Table 5, we can conclude that the DEP IST is a good predictor of success up to
45 months of service. Again, we investigate this relationship further with a logit
regression equation, where the dependent variable in the logit regression equation is
a zero or one indicator variable for attrition in the first 45 months of service. These
logit regression results are shown in Table 16 of the appendix.
From the coefficient estimates in Table 16, we see that both male and female recruits
in the IST middle third or IST bottom third have a higher attrition rate at 45 months
than those in the IST top third. The marginal changes for the two IST category
variables shown in Table 16 imply that male recruits in the IST middle third have a
3.1-percentage-point higher overall 45-month attrition rate, and those in the IST
bottom third have a 7.5-percentage-point higher overall 45-month attrition rate, than
those in the IST top third. The corresponding marginal changes for female recruits
imply that female recruits in the IST middle third have a 5.5-percentage-point higher
overall 45-month attrition rate, and those in the IST bottom third have an 11.4-
percentage-point higher rate, than those in the IST top third.
Relationship between DEP IST score and
conditional 24-month and 45-month attrition
The attrition rates at 24 months of service, given that the recruit completes recruit
training, and the attrition rates at 45 months, given that the recruit completes 24
12
months of service, are known as conditional attrition rates.18 They are shown in Table
6 by IST third.
Table 6. Conditional attrition probabilities to 24 and 45 months by IST category and
gender
Men Women
Recruit
training
attrition
Conditional
probability of
attrition
before 24
months of
service
Conditional
probability
of attrition
before 45
months
Recruit
training
attrition
Conditional
probability
of attrition
before 24
months of
service
Conditional
probability of
attrition before 45
months
IST top
third
4.7% 5.6% 6.3% 9.8% 7.1% 8.6%
IST
middle
third
7.1% 6.5% 7.4% 14.3% 8.9% 9.6%
IST
bottom
third
10.9% 6.9% 8.6% 20.2% 9.9% 9.9%
Overall 7.6% 6.5% 7.8% 14.8% 8.8% 9.7%
Source: CNA Marine Corps personnel files
The relationship between the three DEP IST categories and conditional attrition
continues to be strong and persistent after both 24 and 45 months of service. The
24-month attrition rates, conditional on having completed recruit training, for those
in the IST top third are lower than those for Marines in the IST middle and bottom
thirds. Similarly, 45-month attrition rates, conditional on having completed 24
months of service, for those in the IST top third are lower than those for Marines in
the IST middle and bottom thirds. These relationships hold up equally well for both
men and women.
From Table 6, we can conclude that the DEP IST is a good predictor of success in the
first 24 months of service, given that a Marine completed recruit training, as well as a
good predictor of success between 24 and 45 months of service, given that a Marine
18 To compute the conditional probability we use the rule P(B|A) = P(A, B)/P(A), where P(A, B) is
the joint probability of completing recruit training (event A) and attriting before the completion
of 24 months of service but after recruit training (event B), and P(B|A) is the conditional
probability of B given that A occurred. For our purposes, this conditional probability is the
probability of attrition before the completion of 24 months of service but after recruit training
(event B), given that the recruit completed recruit training (event A). Similarly, we can compute
the conditional probability of attriting before the completion of 45 months of service, given
that a Marine completed 24 months of service.
13
successfully completed 24 months of service. As we did with the other attrition rates,
we investigate the relationships further with a logit regression equation, where the
dependent variable in the logit regression equation is a zero or one indicator variable
for attrition. These conditional logit regression results are shown in Table 17 (for 24
months of service) and Table 18 (for 45 months of service).
From the coefficient estimates in Table 17, we see that both men and women in the
IST middle and bottom thirds have higher conditional 24-month attrition rates than
those in the IST top third. The marginal changes for the two IST category variables
shown in Table 14 imply that male Marines in the IST middle third have a (relatively
small) 0.7-percent-percentage point higher conditional 24-month attrition rate, and
those in the IST bottom third have a 1-percentage-point higher conditional 24-month
attrition rate, than those in the IST top third.
The marginal changes of the two IST category variables in Table 17 also indicate that
female Marines in the IST middle third have a 1.3-percentage-point higher conditional
24-month attrition rate, and those in the bottom IST third have a 2.3-percentage-
point higher conditional 24-month attrition rate, than those in the IST top third.
From the coefficient estimates in Table 18 in the appendix, we see that both men and
women in the IST middle and bottom thirds have a higher conditional 45-month
attrition rate than those in the IST top third, although for women the effect is smaller
than the effect for the conditional 24-month attrition rate, and the coefficient
estimates are not statistically significantly different from zero for women. The
marginal changes for the two IST category variables shown in Table 18 imply that
male recruits in the IST middle third have a (relatively small) 0.8 -percentage-point
higher conditional 45-month attrition rate, and those in the IST bottom third have a
2-percentage-point higher conditional 45-month attrition rate, than those in the IST
top third.
The marginal changes of the two IST category variables in Table 18 also indicate that
female Marines in the IST middle third have a (small) 0.4-percentage-point higher
conditional 45-month attrition rate, and those in the IST bottom third have a (small)
0.7-percentage-point higher conditional 45-month attrition rate, than those in the IST
top third, although the coefficient estimates underlying both of these percentage
attrition impacts are not statistically significantly different from zero.
14
Relationship Between DEP IST Scores
and Recruit Training Injuries
In this section, we discuss the relationship between DEP IST scores and the
probability of being injured while in recruit training. This relationship is summarized
in Table 7 below and in Table 19 in the appendix. We analyze only recruits who
trained at Parris Island, since data were not available for those in San Diego.
We construct a proxy variable for whether a recruit is injured in recruit training using
a combination of MCRISS and MCTIMS data. The recruit is defined as injured if he or
she was discharged for medical reasons from recruit training or was medically
recycled in his or her first attempt at recruit training.
Table 7 shows the relationship between our categories of DEP IST scores and injury
rates in recruit training for both men and women. From Table 7, we see a slight
negative relationship between the three categories of IST scores and injury rate in
recruit training: those who received higher IST scores have lower injury rates. For
men, the injury rate is 2.4 percent for recruits in the IST top third, 3.6 for those in
the IST middle third, and 4.6 percent for those in the IST bottom third. For women,
the injury rates are 4.6 percent for recruits in the IST top third, 5.8 percent for those
in the IST middle third, and 7.6 percent for those in the IST bottom third. The
overall injury rates are 3.6 percent for men and 6.0 percent for women.
Table 7. Recruit training injury rates by DEP IST categories and gender, FY05 –FY13
accessions
Men
Women
Total
recruits Injuries
Total
recruits Injuries
IST top third 33,833 2.4% 5,931 4.6%
IST middle third 34,706 3.6% 5,985 5.8%
IST bottom third 34,770 4.6% 6,058 7.6%
Total 103,309 3.6% 17,974 6.0%
Source: CNA Marine Corps personnel files
15
From Table 7, we can conclude that the DEP IST is a good predictor of injury rates in
recruit training. As we did with our analysis of attrition rates, we further investigate
the relationship between the DEP IST scores and injury rates by conducting a logit
regression equation, where the dependent variable is an indicator variable that takes
on the value of one if the recruit was injured in recruit training, and zero otherwise.
These logit regression results are shown in Table 19 of the appendix.
From the coefficient estimates in Table 19, we see that those in the IST middle or
bottom third have a greater likelihood of being injured in recruit training than those
in the top IST group, for both men and women. The marginal changes for the two
IST category variables shown in Table 19 imply that male recruits in the IST middle
third have a 1.2-percentage-point higher injury rate, and those in the IST bottom
third have a 2.2-percentage-point higher injury rate, than those in the IST top third.
The marginal changes of the two IST category variables in Table 19 for female
recruits imply that female recruits in the IST middle third have a 1.2-percentage-
point higher injury rate, and those in the IST bottom third have a 3.0-percentage-
point higher injury rate, than those in the IST top third.
16
Relationship Between DEP IST Scores
and PFT and CFT Scores
In this section, we discuss the relationship between the DEP IST score and the PFT
and CFT scores attained at the end of recruit training, as well as the relationship
between the DEP IST score and the first and second annual PFT and CFT scores
attained in a Marine’s early career.
Relationship between DEP IST score and final
recruit training PFT score
The relationship between the DEP IST score and the final recruit training PFT score is
summarized in Table 8 below and in Table 20 in the appendix.
Table 8 shows the relationship between the four categories19 of DEP IST scores and
four categories of final recruit training PFT20 scores.21
19 For men who took the final PFT in recruit training, a recruit is classified as being in the
bottom third of the IST if his IST score was 165 or less (IST ≤ 165); if his IST score was higher
than 165 but lower than 210 (166 ≤ IST ≤ 209) he is classified as being in the middle third; and
if his IST score was greater than 209 but less than 300 (210 ≤ IST ≤ 299) he is classified as
being in the top third of the IST, but not 300. For women, these respective cut-off points are:
IST ≤ 158 for the bottom third; 159 ≤ IST ≤ 204 for the middle third; and 205 ≤ IST ≤ 299 for
the top third, but not 300. For both men and women, there is a fourth category for recruits
who had an IST score of 300.
20 For men, a recruit is classified as being in the bottom third of the final recruit training PFT if
his PFT score was 232 or less (PFT ≤ 232); if his PFT score was higher than 232 but lower than
262 (233 ≤ PFT ≤ 261) he is classified as being in the middle third; and if his PFT was greater
than 261 but less than 300 (262 ≤ PFT ≤ 299) he is classified as being in the top third of the
PFT, but not 300. For women, these respective cut-off points are: PFT ≤ 242 for the bottom
third; 243 ≤ PFT ≤ 269 for the middle third; and 270 ≤ PFT ≤ 299 for the top third, but not 300.
For both men and women, there is a fourth category for recruits who had a final recruit
training PFT score of 300.
21 While the average difference between the PFT scores taken at the end of recruit training and
the IST scores taken when the recruit first enters the DEP is an increase of 55 points, for those
17
Table 8. Relationship between DEP IST categories and recruit training PFT
categories by gender, FY05-FY13 accessions
Men Women
Total
recruits
PFT 300
(%)
PFT top
third (but
not 300)
(%)
PFT
middle
third
(%)
PFT
bottom
third
(%)
Total
recruits
PFT
300
(%)
PFT top
third
(but not
300) (%)
PFT
middle
third
(%)
PFT
bottom
third
(%)
IST 300 778 21.7% 71.7% 4.1% 2.4% 85 23.5% 65.9% 9.4% 1.2%
IST top
third
(but not
300)
74,706 1.7% 57.9% 30.7% 9.8% 5,503 2.1% 52.2% 32.3% 13.5%
IST
middle
third
77,550 0.2% 26.2% 41.9% 31.7% 5,624 0.4% 28.3% 38.3% 33.0%
IST
bottom
third
76,602 0.1% 10.9% 28.5% 60.5% 5,612 0.1% 13.1% 30.4% 56.3%
Total 229,636 0.7% 31.5% 33.6% 34.1% 16,824 1.0% 31.3% 33.5% 34.2%
Source: CNA Marine Corps personnel files
We can draw two conclusions from the cross tabulations shown in Table 8 between
DEP IST scores and final PFT scores at the end of recruit training. First, the DEP IST
score is a good predictor of the PFT score attained at the end of recruit training.
Second, a significant number of men and women change categories between the DEP
IST and the final PFT taken at the end of recruit training. For example, 13.1 percent
of the women who scored in the IST bottom third later scored in the PFT top third at
the end of recruit training, and 13.5 percent of the women who scored in the IST top
third later scored in the PFT bottom third at the end of recruit training. Eleven
percent of the men who scored in the IST bottom third later scored in the PFT top
third (including 300) at the end of recruit training, and 9.8 percent of the men who
score in the IST top third later scored in the PFT bottom third at the end of recruit
training. This suggests that any classifications to MOSs that depend on physical
fitness should not be finalized until the end of recruit training.
recruits who went from the top third in the IST score to the bottom third in the PFT score, the
average difference between the PFT and IST scores is a decrease of 13 points.
18
We further investigated the relationship between DEP IST and PFT at the end of
recruit training with a tobit22 regression equation. In the tobit regression, the
dependent variable is the PFT score a recruit receives at the end of recruit training.
The tobit regression allows us to formally test whether the IST categories shown in
Table 8 are good predictors of PFT scores at the end of recruit training, while
controlling for other factors. These tobit regression results are shown in Table 20 of
the appendix.
From Table 20, we see from both the coefficient estimate and the marginal change23
that men who have a 300 score on the DEP IST later score, on average, 27.2 points
higher on the PFT taken at the end of recruit training than similar men who are in the
IST top third (but not 300) category. Men who are in the IST middle third later have,
on average, a final PFT score in recruit training that is 19.6 points lower, and those in
the IST bottom third later have a final PFT score that is 38.3 points lower, than
similar men who are in the IST top third (but not 300) category.
Also in Table 20, we see that the pattern for women is similar to that for men in
terms of the relationship between the DEP IST score and the PFT score attained at the
end of recruit training. Women who have a 300 score on the DEP IST later score, on
average, 23.8 points higher on the PFT taken at the end of recruit training than
similar women who are in the IST top third (but not 300) category. Women in the IST
middle third later have, on average, a final PFT score in recruit training that is 16.2
points lower, and those in the IST bottom third later have a final PFT score that is
33.8 points lower, than similar women who are in the IST top third (but not 300)
category.
Relationship between DEP IST score and first
annual PFT score
The relationship between the DEP IST score and the first annual PFT score is
summarized in Table 9 below and in Table 21 in the appendix.
22 As discussed earlier, the tobit regression is similar to an OLS regression, where the tobit
regression accounts for the fact that observed PFT and CFT scores are censored from above by
a score of 300 and from below by a score of 134.
23 Unlike the logit model, which uses a non-linear equation to predict attrition rates, the tobit
model uses a linear equation to predict PFT scores (if we want to predict scores below 134 and
above 300). Hence, for the tobit model, the derivative is the same as the coefficient, but we
give the derivative column in the tobit results in the appendix for completeness.
19
Table 9 shows the relationship between the same four categories24 of DEP IST scores
previously defined, and four categories25 of the first annual PFT taken in a Marine’s
career. As with the relationship between the DEP IST and the PFT taken at the end of
recruit training, from Table 9 we see a strong positive relationship between DEP IST
scores and first annual PFT scores for both men and women. Men and women who do
well on the DEP IST also tend to do well in the first annual PFT, and men and women
who score poorly on the DEP IST also tend to do poorly on the first annual PFT.
However, we also see from Table 9 that a significant number of men and women
change categories between the DEP IST and the first annual PFT. For example, 12.7
percent of men in the IST bottom third later score in the top third (but not 300) of
the first annual PFT, and 11.4 percent of men in the IST top third (but not 300)
category later score in the bottom third of the first annual PFT. A similar pattern
holds for women, as shown in Table 9. For example, 16.7 percent of women in the IST
bottom third later score in the top third (but not 300) of the first annual PFT, and
15.0 percent of women in the IST top third (but not 300) category later score in the
bottom third of the first annual PFT.
24 We exclude most Marines who accessed in FY13 from this analysis since they have not yet
had a chance to complete one year of service. Since we have a slightly different sample of
Marines who took the first annual PFT than we had for the PFT taken at the end of recruit
training, the cut-off points for the three categories of “IST thirds” are slightly different from
those reported earlier. For men who have had the opportunity to complete one year of service,
a Marine is classified as being in the bottom third of the IST if his IST score was 164 or less (IST
≤ 164); if his IST score was higher than 164 but lower than 208 (165 ≤ IST ≤ 207) he is classified
as being in the middle third; and if his IST was greater than 207 but less than 300 (208 ≤ IST ≤
299) he is classified as being in the top third of the IST, but not 300. For women, these
respective cut-off points are: IST ≤ 157 for the bottom third; 158 ≤ IST ≤ 204 for the middle
third; and 205 ≤ IST ≤ 299 for the top third. For both men and women, there is a fourth
category for those who had an IST score of 300.
25 For men, a Marine is classified as being in the bottom third of the first annual PFT if his PFT
score was 228 or less (PFT ≤ 228); if his PFT score was higher than 228 but lower than 260 (229
≤ PFT ≤ 259) he is classified as being in the middle third; and if his PFT was greater than 259
but less than 300 (260 ≤ PFT ≤ 299) he is classified as being in the top third of the PFT, but not
300. For women, these respective cut-off points are: PFT ≤ 225 for the bottom third; 226 ≤ PFT
≤ 261 for the middle third; and 262 ≤ PFT ≤ 299 for the top third, but not 300. For both men
and women, there is a fourth category for Marines who had a first annual PFT score of 300.
20
Table 9. Relationship between DEP IST categories and first year PFT categories by
gender, FY05-FY12 accessions
Men Women
Total
recruits
PFT 300
(%)
PFT top
third
(but not
300)
(%)
PFT
middle
third
(%)
PFT
bottom
third
(%)
Total
recruits
PFT 300
(%)
PFT top
third (but
not 300)
(%)
PFT
middle
third
(%)
PFT bottom
third
(%)
IST 300 627 18.3% 74.2% 5.7% 1.8% 66 18.2% 69.7% 9.1% 3.0%
IST top
third (but
not 300)
58,749 2.2% 55.2% 31.2% 11.4% 4,222 2.4% 50.0% 32.7% 15.0%
IST
middle
third
58,875 0.4% 27.4% 39.7% 32.5% 4,375 0.6% 28.5% 36.6% 34.3%
IST
bottom
third
61,036 0.2% 12.7% 29.6% 57.6% 4,371 0.2% 16.7% 30.7% 52.4%
Total 179,287 1.0% 31.7% 33.3% 34.0% 13,034 1.1% 31.7% 33.2% 33.9%
Source: CNA Marine Corps personnel files
We further investigate the relationship between the DEP IST and the first annual PFT
with a tobit regression equation, where the dependent variable is the first annual PFT
score that a Marine achieves after recruit training. The tobit regression allows us to
formally test whether the IST categories shown in Table 9 are good predictors of the
first annual PFT score, while controlling for other factors. These tobit regression
results are shown in Table 21 of the appendix.
From Table 21 in the appendix, we see from both the coefficient estimate and the
marginal changes that men who have a 300 score on the DEP IST later score, on
average, 28.8 points higher on the first annual PFT than similar men who are in the
IST top third (but not 300) category. Men in the IST middle third later have, on
average, a first annual PFT score that is 20.6 points lower, and men in the IST bottom
third later have a first annual PFT score that is 40.3 points lower, than similar men
who are in the IST top third (but not 300) category.
Also in Table 21, we see a similar pattern for women. Women who score 300 on the
DEP IST later score, on average, 31.6 points higher on the first annual PFT than
similar women who are in the IST top third (but not 300) category. Women in the IST
middle third have, on average, a first annual PFT score that is 20.9 points lower, and
women in the IST bottom third have a first annual PFT score that is 39.4 points
lower, than similar women who are in the IST top third (but not 300) category.
21
In summary, those who perform well on the DEP IST also perform well on the PFT
after the first year of service.
Relationship between DEP IST score and
second annual PFT score
The relationship between the DEP IST score and the second annual PFT score is
summarized in Table 10 below and in Table 22 in the appendix.
Table 10 shows the cross tabulations between the four categories26 of IST scores and
the four categories27 of the second annual PFT taken in a Marine’s career. As with the
relationship between the DEP IST and the PFTs taken at the end of recruit training
and at the end of the first year of service, there continues to be a strong positive
relationship between DEP IST scores and second annual PFT scores for both men and
women. Men and women who do well on the IST also tend to do well on the second
annual PFT, and those who score poorly on the IST also tend to score poorly on the
second annual PFT.
However, we also see from Table 10 that a significant number of men and women
change categories between the DEP IST and the second annual PFT. For example,
13.7 percent of men in the bottom third of the DEP IST later score in the top third
(but not 300) of the second annual PFT, and 12.0 percent of men in the top third (but
26 We exclude most Marines who accessed after FY11 from this analysis since they have not yet
had a chance to complete two years of service. Since we have a slightly different sample of
Marines who took the second annual PFT than we had for the PFT taken at the end of recruit
training or at the end of the first year of service, the cut-off points for the three categories of
“IST thirds” are slightly different from the IST cut-off points taken at the end of recruit training
and at the end of the first year of service. For men who took the second year PFT, a Marine is
classified as being in the bottom third of the IST score if his IST score was 163 or less (IST ≤
163); if his IST score was higher than 163 but lower than 208 (164 ≤ IST ≤ 207) he is classified
as being in the middle third; and if his IST was greater than 207 but less than 300 (208 ≤ IST ≤
299) he is classified as being in the top third of the IST, but not 300. For women, these
respective cut-off points are: IST ≤ 156 for the bottom third; 157 ≤ IST ≤ 202 for the middle
third; and 203 ≤ IST ≤ 299 for the top third, but not 300. For both men and women, there is a
fourth category for those who had an IST score of 300.
27 For men, a Marine is classified as being in the bottom third of the second annual PFT if his
PFT score was 233 or less (PFT ≤ 233); if his PFT score was higher than 233 but lower than 266
(234 ≤ PFT ≤ 265) he is classified as being in the middle third; and if his PFT was greater than
265 but less than 300 (266 ≤ PFT ≤ 299) he is classified as being in the top third of the PFT, but
not 300. For women, these respective cut-off points are: PFT ≤ 232 for the bottom third; 233 ≤
PFT ≤ 265 for the middle third; and 266 ≤ PFT ≤ 299 for the top third, but not 300. For both
men and women, there is a fourth category for Marines who had a second annual PFT score of
300.
22
not 300) of the DEP IST later score in the bottom third of the second annual PFT. A
similar pattern holds for women in Table 10. For example, 17.3 percent of women in
the IST bottom third later score in the top third (but not 300) of the second annual
PFT, and 15.7 percent of women in the IST top third (but not 300) later score in the
bottom third of the second annual PFT.
Table 10. Relationship between DEP IST categories and second year PFT
categories by gender, FY05-FY11 accessions
Men Women
Total
recruits
PFT 300
(%)
PFT top
third (but
not 300)
(%)
PFT
middle
third
(%)
PFT
bottom
third
(%)
Total
recruits
PFT 300
(%)
PFT top
third (but
not 300)
(%)
PFT
middle
third
(%)
PFT
bottom
third
(%)
IST 300 584 18.8% 70.6% 8.6% 2.1% 63 17.5% 66.7% 15.9% 0.0%
IST top
third (but
not 300)
55,324 2.9% 52.9% 32.2% 12.0% 4,041 2.7% 48.9% 32.7% 15.7%
IST
middle
third
57,418 0.7% 27.8% 38.8% 32.7% 4,125 1.3% 28.4% 36.3% 33.9%
IST
bottom
third
57,730 0.3% 13.7% 29.7% 56.3% 4,128 0.4% 17.3% 29.6% 52.7%
Total 171,056 1.4% 31.3% 33.5% 33.9% 12,357 1.5% 31.6% 32.8% 34.1%
Source: CNA Marine Corps personnel files
We further investigate the relationship between DEP IST and the second annual PFT
with a tobit regression equation, where the dependent variable is the second annual
PFT score that a Marine achieves after recruit training. The tobit regression allows us
to formally test whether the IST categories shown in Table 10 are good predictors of
the second annual PFT score, while controlling for other factors. These tobit
regression results are shown in Table 22 of the appendix.
From Table 22 in the appendix for men, we see from both the coefficient estimate
and the marginal changes that men who have a 300 score on the DEP IST later score,
on average, 24.9 points higher on the second annual PFT than similar men who are in
the IST top third (but not 300) category. Men who are in the IST middle third later
have, on average, a second annual PFT score that is 19.4 points lower, and men who
are in the IST bottom third later have a second annual PFT score that is 38.9 points
lower, than similar men in the IST top third (but not 300) category.
In Table 22, we see that the pattern for women is similar to that for men. Women
who score a 300 on the DEP IST later score, on average, 25.3 points higher on the
second annual PFT than similar women in the IST top third (but not 300) category.
Women in the IST middle third later have, on average, a second annual PFT score that
is 18.9 points lower, and women who are in the IST bottom third later have a second
23
annual PFT score that is 37 points lower, than similar women in the IST top third (but
not 300) category.
In summary, those who perform well on the DEP IST continue to perform well on the
PFT after the second year of service.
Relationship between DEP IST score and final
recruit training CFT score
The relationship between the DEP IST score and the final CFT score at the end of
recruit training is summarized in Table 11 below and in Table 23 in the appendix.
Table 11 shows the cross tabulations between four categories28 of DEP IST scores and
four categories29 of final CFT scores at the end of recruit training. As with the
relationship between the DEP IST and the PFT taken at the end of recruit training,
Table 11 shows a strong positive relationship between DEP IST scores and CFT scores
attained at the end of recruit training for both men and women. Men and women
who do well on the IST also tend to do well on the CFT taken at the end of recruit
training, and men and women who score poorly on the IST also tend to score poorly
on the CFT taken at the end of recruit training.
28 For men who took the final recruit training CFT, a recruit is classified as being in the bottom
third of the IST if his IST score was 176 or less (IST ≤ 176); if his IST score was higher than 176
but lower than 219 (177 ≤ IST ≤ 218) he is classified as being in the middle third; and if his IST
was greater than 218 but less than 300 (219 ≤ IST ≤ 299) he is classified as being in the top
third of the IST, but not 300. For women, these respective cut-off points are: IST ≤ 163 for the
bottom third; 164 ≤ IST ≤ 210 for the middle third; and 211 ≤ IST ≤ 299 for the top third, but
not 300. For both men and women, there is a fourth category for recruits who had an IST score
of 300.
29 For men, a recruit is classified as being in the bottom third of the final recruit training CFT if
his CFT score was 278 or less (CFT ≤ 278); if his CFT score was higher than 278 but lower than
291 (279 ≤ CFT ≤ 290) he is classified as being in the middle third; and if his CFT was greater
than 290 but less than 300 (291 ≤ CFT ≤ 299) he is classified as being in the top third of the
CFT, but not 300. For women, these respective cut-off points are: CFT ≤ 274 for the bottom
third; 275 ≤ CFT ≤ 289 for the middle third; and 290 ≤ CFT ≤ 299 for the top third, but not 300.
For both men and women, there is a fourth category for recruits who had a final recruit
training CFT score of 300.
24
Table 11. Relationship between DEP IST categories and recruit training CFT
categories by gender, FY05-FY13 accessions
Men Women
Total
recruits
CFT 300
(%)
CFT top
third (but
not 300)
(%)
CFT
middle
third
(%)
CFT
bottom
third
(%)
Total
recruits
CFT
300
(%)
CFT top
third (but
not 300)
(%)
CFT
middle
third
(%)
CFT
bottom
third
(%)
IST 300 419 31.7% 36.0% 22.2% 10.0% 44 34.1% 36.4% 15.9% 13.6%
IST top
third
(but not
300)
35,147 11.8% 33.2% 33.5% 21.5% 3,046 11.8% 32.8% 35.3% 20.1%
IST
middle
third
35,841 5.4% 24.8% 37.2% 32.5% 3,101 4.5% 25.4% 36.7% 33.5%
IST
bottom
third
36,019 2.5% 15.7% 34.3% 47.5% 3,096 1.8% 16.6% 33.4% 48.2%
Total 107,426 6.6% 24.6% 35.0% 33.9% 9,287 6.1% 25.0% 35.1% 33.9%
Source: CNA Marine Corps personnel files
However, we also see from Table 11 that a significant number of men and women
change categories between the DEP IST and the CFT taken at the end of recruit
training. For example, 18.2 percent (2.5 percent plus 15.7 percent) of men in the IST
bottom third later either score 300 on the CFT or score in the top third (but not 300)
of the CFT taken at the end of recruit training, and 21.5 percent of men in the IST top
third (but not 300) group score in the bottom third of the CFT taken at the end of
recruit training. Also, 10 percent of men that scored 300 or higher in the IST were in
the bottom third of the CFT taken at the end of recruit training A similar pattern
holds for women, as shown in Table 11. For example, 18.4 percent (1.8 percent plus
16.6 percent) of women in the IST bottom third later either score 300 on the CFT
taken at the end of recruit training or score in the top third (but not 300) of the CFT
taken at the end of recruit training, and 20.1 percent of women in the IST top third
(but not 300) group later score in the bottom third of the recruit training CFT. Also,
13.6 percent of women that scored 300 or higher in the IST were in the bottom third
of the CFT taken at the end of recruit training As we find with the PFT taken at the
end of recruit training, this suggests that any MOS classifications that depend on
physical fitness should not be finalized until the end of recruit training.
We further investigate the relationship between the DEP IST and the CFT in recruit
training with a tobit regression equation. In the tobit regression, the dependent
variable is the CFT score a recruit receives at the end of recruit training. The tobit
regression allows us to formally test whether the IST categories shown in Table 11
25
are good predictors of CFT scores at the end of recruit training, while controlling for
other factors. These tobit regression results are shown in Table 23 of the appendix.
From Table 23 in the appendix, we see that men who score a 300 on the DEP IST will
score, on average, 8.9 points higher on the CFT taken at the end of recruit training
than similar men who are in the IST top third (but not 300) category. Men in the IST
middle third later have, on average, a final CFT score in recruit training that is 5
points lower than similar men in the IST top third (but not 300) group. Finally, men
in the IST bottom third later have a final CFT score that is 10.3 points lower than
similar men who are in the IST top third (but not 300) category.
In Table 23, we see that the pattern for women is similar to that for men. Women
who score a 300 on the DEP IST later score, on average, 7.8 points higher on the CFT
at the end of recruit training than similar women who are in the IST top third (but
not 300) category. Women in the IST middle third later have, on average, a final CFT
score in recruit training that is 5.8 points lower, and women in the IST bottom third,
later have a final PFT score that is 11 points lower, than similar women who are in
the IST top third (but not 300) category.
In summary, comparing the IST category marginal changes in the tobit regression
estimates for both men and women shown in Table 23 for the CFT taken at the end
of recruit training to those in Table 20 for the PFT taken at the end of recruit
training, we see that the effect of the IST category on CFT score at the end of recruit
training is positive, but smaller, than the effect of the IST score on the PFT taken at
the end of recruit training. However, these effects still indicate a positive relationship
between DEP IST and recruit training CFT scores.
Relationship between DEP IST score and first
annual CFT score
The relationship between DEP IST and first annual CFT scores is summarized in
Table 12 below and in Table 24 in the appendix.
26
Table 12 shows the relationship between four categories30 of DEP IST scores and four
categories31 of CFT scores for those taken in the first year of a Marine’s career. As
with the relationship between the DEP IST and the CFT taken at the end of recruit
training, from Table 12 we see a strong positive relationship between IST scores and
first annual CFT scores for both men and women. Men and women who do well on
the IST also tend to do well on the first annual CFT, and men and women who do
poorly on the IST also tend to do poorly on the first annual CFT.
However, we also see from Table 12 that a significant number of men and women
change categories between the DEP IST and the first annual CFT. For example, 19.5
percent (5.4 percent plus 14.1 percent) of men in the IST bottom third later either
score 300 on the first annual CFT or score in the top third of the first annual CFT,
and 20.6 percent of men in the IST top third (but not 300) category score in the
bottom third of the first annual CFT. Also, 10.9 percent of men that scored 300 or
higher in the IST were in the bottom third of the first annual CFT. A similar pattern
holds for women, as indicated by Table 12. For example, 18.2 percent (3.4 percent
plus 14.8 percent) of women in the IST bottom third later score either 300 on the
CFT taken at the end of recruit training or score in the top third (but not 300) of the
CFT taken at the end of recruit training, and 22.8 percent of women in the IST top
third (but not 300) group later score in the bottom third of the first annual CFT.
30 We exclude most Marines who accessed in FY13 from this analysis since they have not yet
had a chance to complete one year of service. Since we have a slightly different sample of
Marines who took the first annual CFT than we had for the CFT taken at the end of recruit
training, the cut-off points for the three categories of “IST thirds” are slightly different from
the IST cut-off points taken at the end of recruit training. For men who took the first annual
CFT, a Marine is classified as being in the bottom third of the IST if his IST score was 170 or
less (IST ≤ 170); if his IST score was higher than 170 but lower than 214 (171 ≤ IST ≤ 213) he is
classified as being in the middle third; and if his IST was greater than 213 but less than 300
(214 ≤ IST ≥ 299) he is classified as being in the top third of IST, but not 300. For women,
these respective cut-off points are: IST ≤ 162 for the bottom third; 163 ≤ IST ≤ 208 for the
middle third; and 209 ≤ IST ≤ 299 for the top third, but not 300. For both men and women,
there is a fourth category for those who had an IST score of 300.
31 For men, a Marine is classified as being in the bottom third of the first annual CFT if his CFT
score was 279 or less (CFT ≤ 279); if his CFT score was higher than 279 but lower than 293 (280
≤ CFT ≤ 292) he is classified as being in the middle third; and if his CFT was greater than 292
but less than 300 (293 ≤ CFT ≤ 299) he is classified as being in the top third of the CFT, but not
300. For women, these respective cut-off points are: CFT ≤ 277 for the bottom third; 278 ≤ CFT
≤ 291 for the middle third; and 292 ≤ CFT ≤ 299 for the top third, but not 300. For both men
and women, there is a fourth category for Marines who had a first annual CFT score of 300.
27
Table 12. Relationship between DEP IST categories and first year CFT categories by
gender, FY05-FY12 accessions
Men Women
Total
recruits
CFT 300
(%)
CFT top
third (but
not 300)
(%)
CFT
middle
third
(%)
CFT
bottom
third
(%)
Total
recruits
CFT
300
(%)
CFT top
third (but
not 300)
(%)
CFT
middle
third
(%)
CFT
bottom
third
(%)
IST 300 449 41.9% 27.8% 19.4% 10.9% 54 31.5% 29.6% 25.9% 13.0%
IST top
third
(but not
300)
39,730 20.7% 26.4% 32.3% 20.6% 3,087 17.1% 29.0% 31.1% 22.8%
IST
middle
third
40,521 10.8% 21.2% 35.8% 32.3% 3,071 7.7% 23.5% 35.9% 32.9%
IST
bottom
third
40,429 5.4% 14.1% 33.2% 47.3% 3,220 3.4% 14.8% 34.5% 47.4%
Total 121,129 12.4% 20.6% 33.7% 33.4% 9,432 9.4% 22.4% 33.8% 34.4%
Source: CNA Marine Corps personnel files
We investigate the relationship between DEP IST and the first annual CFT further
with a tobit regression equation, where the dependent variable is the first annual CFT
score that a Marine achieves after recruit training. The tobit regression allows us to
formally test whether the IST categories shown in Table 12 are good predictors of the
first annual CFT score, while controlling for other factors. These tobit regression
results are shown in Table 24 of the appendix.
From Table 24, we see from both the coefficient estimate and the marginal change
that men who score a 300 on the DEP IST later score, on average, 12.2 points higher
on the first annual CFT than similar men who are in the IST top third (but not 300)
category. Men in the IST middle third later have, on average, a first annual CFT score
that is 6.7 points lower, and those in the DEP IST bottom third later have a score that
is 12.5 points lower, than similar men who are in the IST top third (but not 300)
category.
In Table 24, we see that the pattern for women is similar to that for men. Women
who score 300 on the DEP IST later score, on average, 14.1 points higher on the first
annual CFT than similar women in the IST top third (but not 300) category. Women
28
in the IST middle third later have, on average, a first annual CFT score that is 6.5
points lower, and those in the bottom third later have a first annual CFT score that is
13.0 points lower, than similar women in the IST top third (but not 300) category.
In summary, those who perform well on the DEP IST continue to perform well on the
CFT after the first year of service.
Relationship between DEP IST score and
second annual CFT score
The relationship between the DEP IST categories32 and the second annual CFT score33
is summarized in Table 13 below and in Table 25 in the appendix. Table 13 shows
the relationship between DEP IST and CFT scores in the second year of a Marine’s
career. As with the relationship between the DEP IST and the CFTs taken at the end of
recruit training and at the end of the first year of service, there continues to be a
strong positive relationship between IST scores and CFT scores for both men and
women. Men and women who do well on the DEP IST also tend to do well on the
second annual CFT, and those who do poorly on the DEP IST also tend to do poorly
on the second annual CFT.
32 We exclude most Marines who accessed after FY11 from this analysis since they have not yet
had a chance to complete two years of service. Since we have a slightly different sample of
Marines who took the first annual CFT than we had for the CFTs taken at the end of recruit
training or at the end of the first year of service, the cut-off points for the four categories of
“IST thirds” are slightly different from those previously reported. For men who took the
second annual CFT, a Marine is classified as being in the bottom third of IST if his IST score
was 166 or less (IST ≤ 166); if his IST score was higher than 166 but lower than 210 (167 ≤ IST
≤ 209) he is classified as being in the middle third; and if his IST was greater than 209 but less
than 300 ( 210 ≤ IST ≥ 299) he is classified as being in the top third of the IST, but not 300. For
women, these respective cut-off points are: IST ≤ 157 for the bottom third; 158 ≤ IST ≤ 203 for
the middle third; and 204 ≤ IST ≤ 299 for the top third, but not 300. For both male and female
Marines, there is a fourth category for those who had an IST score of 300.
33 For men, a Marine is classified as being in the bottom third of the second annual CFT if his
CFT score was 279 or less (CFT ≤ 279); if his CFT score was higher than 279 but lower than 294
(280 ≤ CFT ≤ 293) he is classified as being in the middle third; and if his CFT was greater than
293 but less than 300 (294 ≤ CFT ≤ 299) he is classified as being in the top third of the CFT, but
not 300. For women, these respective cut-off points are: CFT ≤ 277 for the bottom third; 278 ≤
CFT ≤ 291 for the middle third; and 292 ≤ CFT ≤ 299 for the top third, but not 300. For both
men and women, there is a fourth category for Marines who had a second annual CFT score of
300.
29
Table 13. Relationship between DEP IST categories and second year CFT categories
by gender, FY05-FY11 accessions
Men Women
Total
recruits
CFT 300
(%)
CFT top
third (but
not 300)
(%)
CFT
middle
third
(%)
CFT
bottom
third
(%)
Total
recruits
CFT
300
(%)
CFT top
third (but
not 300)
(%)
CFT
middle
third
(%)
CFT
bottom
third
(%)
IST 300 395 41.8% 23.8% 25.1% 9.4% 43 39.5% 20.9% 27.9% 11.6%
IST top
third
(but not
300)
37,666 22.8% 21.8% 35.0% 20.4% 2,707 18.5% 27.3% 30.7% 23.6%
IST
middle
third
38,065 12.5% 17.3% 38.1% 32.1% 2,761 9.0% 22.9% 35.0% 33.1%
IST
bottom
third
39,167 6.2% 11.2% 34.4% 48.2% 2,808 4.8% 14.9% 33.8% 46.5%
Total 115,293 13.8% 16.7% 35.8% 33.7% 8,319 10.8% 21.6% 33.2% 34.4%
Source: CNA Marine Corps personnel files
However, we also see from Table 13 that a significant number of men and women
change categories between the DEP IST and the second annual CFT. For example,
17.4 percent (6.2 percent plus 11.2 percent) of men in the IST bottom third later
either score 300 or score in the top third (but not 300) of the second annual CFT, and
20.4 percent of men in the IST top third (but not 300) group later score in the bottom
third of the second annual CFT. Also, 9.4 percent of men that scored 300 or higher
in the IST were in the bottom third of the second annual CFT A similar pattern holds
for women, as shown in Table 13. For example, 19.7 percent (4.8 percent plus 14.9
percent) of women in the DEP IST bottom third later either score 300 or score in the
top third (but not 300) of the second annual CFT, and 23.6 percent of women in the
IST top third (but not 300) group later score in the bottom third of the second annual
CFT. Also, 11.6 percent of women that scored 300 or higher in the IST were in the
bottom third of the second annual CFT.
We further investigate the relationship between DEP IST and the second annual CFT
by using a tobit regression equation, where the dependent variable is the second
annual CFT score that a Marine achieves after recruit training. The tobit regression
allows us to formally test whether the IST categories shown in Table 13 are good
predictors of the second annual CFT score, while controlling for other factors. These
tobit regression results are shown in Table 25 in the appendix.
30
From Table 25, we see that men who score 300 on the DEP IST later score, on
average, 13.2 points higher on the second annual CFT than similar men who are in
the IST top third (but not 300) category. Men in the IST middle third have, on
average, a second annual PFT score that is 7.0 points lower, and men in the IST
bottom third have a score that is 14.5 points lower, than similar men in the IST top
third (but not 300) category.
In Table 25, we see that the pattern for women is similar to that for men. Women
who score 300 on the DEP IST later score, on average, 15.9 points higher on the
second annual CFT than similar women in the IST top third (but not 300) category.
Women in the IST middle third later have, on average, a second annual CFT score that
is 7.7 points lower, and women in the IST bottom third later have a score that is 15.9
points lower, than similar women in the IST top third (but not 300) category.
In summary, those who perform well on the DEP IST tend to perform well on the CFT
after the second year of service.
31
Conclusions
We find that the DEP IST score is a good predictor of recruit training attrition, injury
rates, and PFT and CFT scores: a higher DEP IST score appears to lead to both lower
attrition and injury rates, and to higher PFT and CFT scores early in a Marine’s
career. The effect of IST scores on conditional 24-month and 45-month attrition was
less pronounced than the effect of DEP IST scores on recruit training attrition, but
our results still predict lower conditional 24- and 45-month attrition rates for both
men and women who scored high on the DEP IST.
The relative effect of DEP IST scores on recruit training attrition is the same for men
and women. Male recruits in the IST top third have a 4.7-percent attrition rate in
recruit training, whereas those in the bottom third have a 10.9-percent attrition rate
in recruit training. Female recruits in the IST top third have a 9.8-percent attrition
rate in recruit training, whereas female recruits in the IST bottom third have a 20.2-
percent attrition rate in recruit training. In short, attrition rates for both men and
women in the IST bottom third are roughly twice those for men and women in the
IST top third.
Based on a logit regression, the effect of the DEP IST on conditional 24-month and
45-month attrition rates is considerably smaller than similar logit regression
estimates for recruit training attrition. Men in the IST bottom third have a 5.2.-
percentage-point higher rate of attrition from recruit training than those in the IST
top third. Women in the IST bottom third have a 9.6-percentage-point higher rate of
attrition from recruit training than those in the IST top third. In terms of conditional
24-month attrition, men in the IST bottom third have a 1.0-percentage-point higher
attrition rate than those in the IST top third. In that same category, women in the IST
bottom third have a 2.3-percentage-point higher attrition rate than women in the IST
top third.
When analyzing the relationship between DEP IST scores and the probability of being
injured while in recruit training, we find that injury rates are lower for both men and
women who have higher IST scores. We find that although the average recruit
training injury rate for men is 3.6 percent, men in the IST top third have only a 2.5-
percent predicted injury rate, men in the IST middle third have a 3.7-percent
predicted injury rate, and men in the IST bottom third have a 4.7-percent predicted
injury rate in recruit training.
32
We see a similar relationship between DEP IST scores and recruit training injury rates
for women. Although the average injury rate for women is 6.0 percent, women in the
IST top third have only a 4.7-percent predicted injury rate, women in the IST middle
third have a 5.9-percent predicted injury rate, and women in the IST bottom third
have a 7.7-percent predicted injury rate. We therefore conclude that higher DEP IST
scores lead to lower predicted recruit training injury rates for both men and women.
Our analysis of the relationship between DEP IST scores and PFT and CFT scores
taken during the early part of a Marine’s career indicates that higher DEP IST scores
lead to higher PFT and CFT scores. These effects are smaller for the CFT than for the
PFT, and for the CFT the effects decline over the course of a Marine’s first two years
of service.
We caution that, although the DEP IST is a reliable predictor of PFT and CFT scores
attained in the first two years of a Marine’s initial enlistment, there is a significant
percentage of recruits (13.2 percent of women for the PFT and 18.2 percent of
women for the CFT) who score in the bottom third of the DEP IST and then score
either 300 or in the top third on the PFT and CFT in recruit training. Similarly, there
is a significant percentage of recruits (14.7 percent of women for the PFT and 35.8
percent for the CFT) who score either 300 or in the top third on the DEP IST and then
score in the bottom third on the PFT and CFT in recruit training.
We recommend that the development of any policy to use DEP physical fitness
indicators to support classification decisions include two provisions. First, recruits
who were denied a physically demanding PEF because they attained a low DEP IST
score should be eligible for the physically demanding PEF if they subsequently meet
the physical fitness standard for that PEF at the end of recruit training. Second,
recruits who were given a physically demanding PEF because they attained a high
DEP IST score should be classified into a less physically demanding PEF if they fail to
meet the physical fitness standard for the physically demanding PEF at the end of
recruit training.
33
Appendix: Logit and Tobit Regression
Results
Tables 14–26 present coefficient estimates of the logit and tobit regressions
discussed in the main text.
Table 14. Recruit training attrition logit regression estimates by gender, FY05-FY13
accessions
Men
Women
Mean Coeff Derivative
Mean Coeff Derivative
(z)
(z)
IST middle third
0.336 0.364** 0.0204
0.334 0.391** 0.0407
[17.17]
[7.46]
IST bottom third
0.336 0.772** 0.0519
0.336 0.793** 0.0956
[38.13]
[15.83]
High quality
0.643 -0.239** -0.0168
0.617 -0.179** -0.0220
[15.26]
[4.40]
DEP months >=3
0.694 -0.123** -0.0086
0.610 -0.149** -0.0183
[6.99]
[3.53]
No enlistment waivers
0.504 -0.131** -0.0090
0.511 -0.170** -0.0207
[8.32]
[4.28]
Age at accession
19.354 0.031** 0.0022
19.282 -0.007 -0.0008
[8.41]
[0.67]
Enlistment bonus
0.176 -0.082** -0.0055
0.214 -0.013 -0.0016
34
Men
Women
Mean Coeff Derivative
Mean Coeff Derivative
(z)
(z)
[3.88]
[0.27]
Accession trimester
Oct-Jan accession
0.300 0.134** 0.0090
0.315 0.187** 0.0221
[7.37]
[3.90]
Feb-Apr accession
0.211 0.210** 0.0146
0.275 0.265** 0.0322
[10.38]
[5.23]
Racial/ethnic
background
Asian
0.033 -0.404** -0.0238
0.040 -0.221* -0.0258
[8.30]
[2.12]
Black
0.091 -0.02 -0.0014
0.152 -0.190** -0.0224
[0.76]
[3.39]
Hispanic
0.176 -0.393** -0.0242
0.235 -0.550** -0.0603
[17.37]
[10.50]
Parris Island
0.484 0.121** 0.0083
N/A N/A N/A
[7.84]
N/A
Constant
-3.854**
-2.234**
[46.11]
[10.55]
Number of observations
257,385
21,910
Chi square (df = 23 & 22)
5,064.21
703.37
Absolute value of z statistics in brackets: + significant at 10%; * significant at 5%; **
significant at 1%. Note: These regressions also controlled for accession FY and for
observations missing race information.
35
Table 15. 24-month attrition logit regression estimates by gender, FY05-FY12
accessions
Men
Women
Mean Coeff Derivative
Mean Coeff Derivative
(z)
(z)
IST middle third
0.334 0.237** 0.0244
0.333 0.337** 0.0510
[13.51]
[6.83]
IST bottom third
0.359 0.503** 0.0575
0.353 0.661** 0.1097
[29.70]
[13.90]
High quality
0.629 -0.236** -0.0277
0.609 -0.113** -0.0190
[17.46]
[2.89]
DEP months >=3
0.665 -0.226** -0.0267
0.585 -0.196** -0.0330
[15.06]
[4.81]
No enlistment waivers
0.484 -0.165** -0.0190
0.501 -0.211** -0.0353
[12.21]
[5.53]
Age at accession
19.399 -0.005 -0.0005
19.293 -0.015 -0.0025
[1.39]
[1.60]
Enlistment bonus
0.200 -0.033+ -0.0037
0.241 -0.052 -0.0086
[1.87]
[1.11]
Accession trimester
Oct-Jan accession
0.332 0.116** 0.0133
0.349 0.190** 0.0314
[7.36]
[4.08]
Feb-Apr accession
0.199 0.105** 0.0120
0.252 0.199** 0.0330
[5.80]
[3.97]
Racial/ethnic
background
36
Men
Women
Mean Coeff Derivative
Mean Coeff Derivative
(z)
(z)
Asian
0.033 -0.387** -0.0395
0.039 -0.373** -0.0585
[9.37]
[3.61]
Black
0.088 -0.094** -0.0106
0.152 -0.268** -0.0433
[4.02]
[4.93]
Hispanic
0.170 -0.376** -0.0397
0.223 -0.599** -0.0914
[19.51]
[11.90]
Parris Island
0.481 0.119** 0.0137
N/A N/A N/A
[8.91]
N/A
Constant
-2.046**
-1.215**
[24.70]
[5.52]
Number of observations 207,351 17,298
Chi square (df = 22 & 21)
3,594.44
618.29
Absolute value of z statistics in brackets: + significant at 10%; * significant at 5%; **
significant at 1%. Note: the regressions also controlled for accession FY and for observations
missing race information.
37
Table 16. 45-month attrition logit regression estimates by gender, FY05-FY10
accessions
Men
Women
Mean Coeff Derivative
Mean Coeff Derivative
(z)
(z)
IST middle third
0.330 0.215** 0.0311
0.331 0.287** 0.0546
[12.42]
[5.54]
IST bottom third
0.395 0.479** 0.0751
0.379 0.568** 0.1143
[28.93]
[11.40]
High quality
0.606 -0.251** -0.0404
0.595 -0.105* -0.0213
[19.24]
[2.55]
DEP months >=3
0.601 -0.212** -0.0341
0.539 -0.193** -0.0392
[14.85]
[4.57]
No enlistment waivers
0.458 -0.232** -0.0367
0.487 -0.208** -0.0422
[17.61]
[5.19]
Age at accession
19.428 -0.016** -0.0026
19.321 -0.017+ -0.0033
[4.94]
[1.69]
Enlistment bonus
0.228 -0.047** -0.0073
0.285 -0.006 -0.0012
[2.88]
[0.13]
Accession trimester
Oct-Jan accession
0.339 0.098** 0.0155
0.346 0.143** 0.0287
[6.19]
[2.91]
Feb-Apr accession
0.219 0.098** 0.0155
0.264 0.165** 0.0332
[5.55]
[3.10]
Racial/ethnic
background
38
Men
Women
Mean Coeff Derivative
Mean Coeff Derivative
(z)
(z)
Asian
0.032 -0.386** -0.0550
0.040 -0.422** -0.0808
[9.64]
[3.93]
Black
0.086 0.027 0.0044
0.149 -0.296** -0.0583
[1.23]
[5.17]
Hispanic
0.164 -0.357** -0.0529
0.212 -0.645** -0.1210
[19.19]
[12.15]
Parris Island
0.480 0.128** 0.0204
N/A N/A N/A
[9.92]
N/A
Constant
-1.052**
-0.565**
[14.06]
[2.66]
Number of observations 159,613 12,925
Chi square (df = 20 & 19)
3,141.71
436.37
Absolute value of z statistics in brackets: + significant at 10%; * significant at 5%; **
significant at 1%. Note: These regressions also controlled for accession FY and for
observations missing race information.
39
Table 17. 24-month conditional attrition logit regression estimates by gender, FY05-
FY12 accessions
Men
Women
Mean Coeff Derivative
Mean Coeff Derivative
(z)
(z)
IST middle third
0.336 0.121** 0.0073 0.336 0.173* 0.0130
[5.13] [2.29]
IST bottom third
0.345 0.165** 0.0101 0.330 0.304** 0.0234
[6.96] [4.06]
High quality
0.635 -0.240** -0.0153 0.617 -0.014 -0.0011
[12.48] [0.21]
DEP months >=3
0.672 -0.326** -0.0211 0.597 -0.290** -0.0230
[15.23] [4.43]
No enlistment waivers
0.490 -0.210** -0.0130 0.512 -0.224** -0.0174
[10.91] [3.65]
Age at accession
19.371 -0.050** -0.0030 19.275 -0.032* -0.0024
[9.53] [2.04]
Enlistment bonus
0.202 0.002 0.0002 0.241 -0.071 -0.0054
[0.10] [0.94]
Accession trimester
Oct-Jan accession
0.330 0.043+ 0.0027 0.346 0.045 0.0035
[1.91] [0.61]
Feb-Apr accession
0.195 -0.025 -0.0015 0.243 0.052 0.0040
[0.95] [0.64]
Racial/ethnic
background
40
Men
Women
Mean Coeff Derivative
Mean Coeff Derivative
(z)
(z)
Asian
0.0337 -0.395** -0.0213 0.039 -0.912** -0.0544
[6.66] [4.30]
Black
0.088 -0.151** -0.0090 0.152 -0.507** -0.0353
[4.46] [5.39]
Hispanic
0.174 -0.348** -0.0120 0.236 -0.611** -0.0414
[12.80] [7.47]
Parris Island
0.478 -0.011 -0.0007 N/A N/A N/A
[0.58] N/A
Constant
-1.397** -1.462**
[11.41] [4.10]
Number of observations 190,026 14,530
Chi square (df = 22 & 21)
1,248.07
238.33
Absolute value of z statistics in brackets: + significant at 10%; * significant at 5%; **
significant at 1%. Note: These regressions also controlled for accession FY and for
observations missing race information.
41
Table 18. 45-month conditional attrition logit regression estimates by gender, FY05-
FY10 accessions
Men
Women
Mean Coeff Derivative
Mean Coeff Derivative
(z)
(z)
IST middle third
0.332 0.140** 0.0081 0.334 0.054 0.0037
[4.81] [0.56]
IST bottom third
0.378 0.315** 0.0197 0.349 0.103 0.0071
[11.24] [1.07]
High quality
0.617 -0.226** -0.0146 0.602 -0.119 -0.0084
[10.06] [1.46]
DEP months >=3
0.615 -0.172** -0.0110 0.558 -0.116 -0.0081
[6.91] [1.39]
No enlistment waivers
0.470 -0.336** -0.0209 0.507 -0.092 -0.0064
[14.69] [1.17]
Age at accession
19.397 -0.055** -0.0034 19.302 -0.021 -0.0014
[8.97] [1.07]
Enlistment bonus
0.231 -0.059* -0.0037 0.289 0.099 0.0070
[2.16] [1.12]
Accession trimester
Oct-Jan accession
0.335 -0.024 -0.0015 0.343 0.086 0.0059
[0.85] [0.88]
Feb-Apr accession
0.215 0.038 0.0024 0.252 0.057 0.0039
[1.24] [0.54]
Racial/ethnic
background
42
Men
Women
Mean Coeff Derivative
Mean Coeff Derivative
(z)
(z)
Asian
0.033 -0.344** -0.0187 0.042 -0.598* -0.0350
[4.90] [2.55]
Black
0.086 0.216** 0.0148 0.153 -0.385** -0.0245
[6.09] [3.28]
Hispanic
0.170 -0.256** -0.0151 0.233 -0.669** -0.0400
[8.13] [6.17]
Parris Island
0.475 0.096** 0.0060 N/A N/A N/A
[4.27] N/A
Constant
-1.138** -1.723**
[8.44] [4.11]
Number of observations 134,880 9,676
Chi square (df = 20 & 19)
1,074.35
101.98
Absolute value of z statistics in brackets: + significant at 10%; * significant at 5%; **
significant at 1%. Note: These regressions also controlled for accession FY and for
observations missing race information.
43
Table 19. Recruit training injury logit regression estimates by gender, FY05-FY13
accessions
Men
Women
Mean Coeff Derivative
Mean Coeff Derivative
(z)
(z)
IST middle third
0.354 0.426** 0.0124 0.337 0.237** 0.0117
[9.52] [2.91]
IST bottom third
0.305 0.672** 0.0222 0.313 0.529** 0.0296
[14.99] [6.65]
High quality
0.646 -0.083* -0.0028 0.624 -0.088 -0.0049
[2.30] [1.31]
DEP months >=3
0.708 -0.045 -0.0015 0.620 -0.102 -0.0057
[1.08] [1.47]
No enlistment waivers
0.476 0.025 0.0008 0.519 -0.015 -0.0009
[0.70] [0.24]
Age at accession
19.426 0.050** 0.0017 19.257 0.052** 0.0029
[6.12] [3.35]
Enlistment bonus
0.169 -0.08 -0.0026 0.209 -0.043 -0.0024
[1.48] [0.48]
Accession trimester
Oct-Jan accession
0.285 -0.112** -0.0035 0.307 0.008 0.0004
[2.62] [0.09]
Feb-Apr accession
0.208 0.253** 0.0094 0.284 0.279** 0.0161
[5.86] [3.48]
Racial/ethnic
background
44
Men
Women
Mean Coeff Derivative
Mean Coeff Derivative
(z)
(z)
Asian
0.022 0.067 0.0023 0.039 -0.201 -0.0104
[0.61] [1.16]
Black
0.138 -0.003 -0.0001 0.151 -0.171+ -0.0090
[0.06] [1.82]
Hispanic
0.119 -0.163** -0.0053 0.242 -0.296** -0.0154
[2.93] [3.69]
Constant
-4.407** -3.404**
[24.97] [10.41]
Number of observations 103,127 17,933
Chi square (df = 22 & 22)
995.29
303.34
Absolute value of z statistics in brackets: + significant at 10%; * significant at 5%; **
significant at 1%. Note: These regressions also controlled for accession FY and for
observations missing race information.
45
Table 20. Recruit training PFT tobit regression estimates by gender, FY05-FY13
accessions
Men
Women
Mean Coeff Derivative
Mean Coeff Derivative
(z)
(z)
IST score of 300
0.003 27.203** 27.2033 0.005 23.781** 23.7814
[28.26] [7.42]
IST middle third
0.338 -19.579** -19.5789 0.334 -16.193** -16.1926
[147.17] [30.16]
IST bottom third
0.334 -38.318** -38.3179 0.334 -33.812** -33.8123
[281.34] [62.51]
High quality
0.649 0.209+ 0.2086 0.631 1.541** 1.5405
[1.80] [3.32]
DEP months >=3
0.703 -3.642** -3.6417 0.625 0.068 0.0677
[27.67] [0.14]
No enlistment waivers
0.512 -0.703** -0.7034 0.523 -0.327 -0.3268
[6.31] [0.73]
Age at accession
19.322 0.372** 0.3723 19.248 0.493** 0.4931
[12.48] [4.30]
Enlistment bonus
0.176 0.853** 0.8528 0.201 1.287* 1.2868
[5.65] [2.20]
Accession trimester
Oct-Jan accession
0.297 -0.858** -0.8583 0.311 0.587 0.5867
[6.59] [1.10]
Feb-Apr accession
0.208 1.338** 1.3379 0.274 0.581 0.5808
46
Men
Women
Mean Coeff Derivative
Mean Coeff Derivative
(z)
(z)
[8.96] [1.02]
Racial/ethnic
background
Asian
0.034 2.795** 2.7950 0.040 -0.991 -0.9912
[9.38] [0.89]
Black
0.090 8.499** 8.4986 0.150 2.850** 2.8500
[0.36] [2.17]
Hispanic
0.182 4.542** 4.5420 0.249 2.108** 2.1078
[44.16] [4.53]
Parris Island
0.479 -5.287** -5.2871 N/A N/A N/A
[47.62] N/A
Constant
268.891** 261.921**
[426.25] [108.92]
Number of observations 229,635 16,824
Chi square (df = 24 & 23)
92,275.64
4,483.98
Absolute value of z statistics in brackets: + significant at 10%; * significant at 5%; **
significant at 1%. Note: These regressions also controlled for accession FY and for
observations missing race information.
47
Table 21. First year PFT tobit regression estimates by gender, FY05-FY12 accessions
Men
Women
Mean Coeff Derivative
Mean Coeff Derivative
(z)
(z)
IST score of 300
0.003 28.805** 28.8055 0.005 31.594** 31.5939
[22.09] [5.98]
IST middle third
0.328 -20.584** -20.5842 0.336 -20.937** -20.9368
[110.87] [23.30]
IST bottom third
0.340 -40.297** -40.2970 0.335 -39.409** -39.4086
[213.97] [43.39]
High quality
0.646 0.311+ 0.3107 0.624 1.102 1.1021
[1.94] [1.42]
DEP months >=3
0.686 -4.923** -4.9232 0.619 -1.485+ -1.4854
[27.18] [1.82]
No enlistment waivers
0.502 0.21 0.2096 0.519 -0.793 -0.7929
[1.35] [1.06]
Age at accession
19.313 0.009 0.0094 19.237 0.963** 0.9635
[0.23] [5.13]
Enlistment bonus
0.191 1.885** 1.8847 0.227 1.850* 1.8499
[9.27] [1.96]
Accession trimester
Oct-Jan accession
0.236 3.784** 3.7845 0.235 4.653** 4.6526
[19.46] [4.89]
48
Men
Women
Mean Coeff Derivative
Mean Coeff Derivative
(z)
(z)
Feb-Apr accession
0.225 4.635** 4.6351 0.288 2.058* 2.0577
[23.12] [2.23]
Racial/ethnic
background
Asian
0.035 3.693** 3.6934 0.042 4.843** 4.8430
[9.01] [2.65]
Black
0.088 8.340** 8.3396 0.155 9.414** 9.4145
[1.70] [1.64]
Hispanic
0.179 4.554** 4.5545 0.247 6.510** 6.5103
[30.90] [9.05]
Parris Island
0.475 -2.985** -2.9855 N/A N/A N/A
[19.37] N/A
Constant
264.492** 238.529**
[261.81] [51.05]
Number of observations 179,287 13,034
Chi square (df = 24 & 23) 49,299.2 2,240.34
Absolute value of z statistics in brackets: + significant at 10%; * significant at 5%; **
significant at 1%. Note: These regressions also controlled for accession FY and for
observations missing race information.
49
Table 22. Second year PFT tobit regression estimates by gender, FY05-FY11
accessions
Men
Women
Mean Coeff Derivative
Mean Coeff Derivative
(z)
(z)
IST score of 300
0.003 24.851** 24.8507 0.005 25.290** 25.2899
[17.75] [4.88]
IST middle third
0.336 -19.405** -19.4054 0.334 -18.857** -18.8574
[98.69] [21.27]
IST bottom third
0.337 -38.865** -38.8647 0.334 -37.035** -37.0348
[193.42] [41.39]
High quality
0.644 0.570** 0.5697 0.622 0.739 0.7389
[3.35] [0.97]
DEP months >=3
0.684 -5.274** -5.2736 0.611 -3.021** -3.0207
[27.48] [3.77]
No enlistment waivers
0.498 0.982** 0.9817 0.521 -1.707* -1.7070
[5.98] [2.32]
Age at accession
19.363 -0.080+ -0.0800 19.262 0.579** 0.5790
[1.86] [3.15]
Enlistment bonus
0.198 2.100** 2.1000 0.235 2.245* 2.2446
[9.91] [2.45]
Accession trimester
Oct-Jan accession
0.322 -0.01 -0.0098 0.335 -1.317 -1.3166
50
Men
Women
Mean Coeff Derivative
Mean Coeff Derivative
(z)
(z)
[0.05] [1.51]
Feb-Apr accession
0.193 1.767** 1.7671 0.242 0.093 0.0931
[7.78] [0.10]
Racial/ethnic
background
Asian
0.035 4.485** 4.4850 0.042 5.851** 5.8506
[10.30] [3.25]
Black
0.088 6.803** 6.8034 0.157 5.440** 5.4402
[0.53] [0.26]
Hispanic
0.178 4.806** 4.8064 0.250 7.177** 7.1766
[23.68] [5.32]
Parris Island
0.475 -3.717** -3.7174 N/A N/A N/A
[22.74] N/A
Constant
278.626** 260.288**
[175.43] [40.44]
Number of observations
171,055 12,357
Chi square (df = 24 & 23)
41,236.9
2,013.07
Absolute value of z statistics in brackets: + significant at 10%; * significant at 5%; **
significant at 1%. Note: These regressions also controlled for accession FY and for
observations missing race information.
51
Table 23. Recruit training CFT tobit regression estimates by gender, FY05-FY13
accessions
Men
Women
Mean Coeff Derivative
Mean Coeff Derivative
(z)
(z)
IST score of 300
0.004 8.906** 8.9061 0.005 7.809** 7.8086
[11.92] [2.95]
IST middle third
0.334 -4.999** -4.9990 0.334 -5.792** -5.7922
[46.41] [13.59]
IST bottom third
0.335 -10.294** -10.2940 0.333 -11.031** -11.0307
[96.22] [25.76]
High quality
0.700 0.177+ 0.1769 0.654 -0.018 -0.0176
[1.84] [0.05]
DEP months >=3
0.852 -1.024** -1.0241 0.679 -0.26 -0.2596
[8.25] [0.65]
No enlistment waivers
0.577 -0.529** -0.5286 0.548 0.382 0.3821
[5.91] [1.08]
Age at accession
19.235 0.277** 0.2773 19.238 0.346** 0.3464
[11.19] [3.73]
Enlistment bonus
0.110 0.08 0.0803 0.131 -0.208 -0.2079
[0.57] [0.40]
Accession trimester
Oct-Jan accession
0.271 -2.157** -2.1570 0.297 -0.816+ -0.8165
52
Men
Women
Mean Coeff Derivative
Mean Coeff Derivative
(z)
(z)
[20.15] [1.86]
Feb-Apr accession
0.177 -0.789** -0.7886 0.264 -2.235** -2.2349
[6.44] [4.89]
Racial/ethnic
background
Asian
0.036 -3.666** -3.6658 0.041 -2.842** -2.8415
[15.47] [3.24]
Black
0.097 -0.901** -0.9014 0.157 -2.810** -2.8099
[5.16] [1.37]
Hispanic
0.197 -1.409** -1.4095 0.269 -2.450** -2.4505
[5.97] [5.71]
Parris Island
0.489 0.192* 0.1916 N/A N/A N/A
[2.15] N/A
Constant
288.332** 281.326**
[559.91] [146.98]
Number of observations 107,425 9,287
Chi square (df = 21 & 19)
27,365.95
1,115.17
Absolute value of z statistics in brackets: + significant at 10%; * significant at 5%; **
significant at 1%. Note: These regressions also controlled for accession FY and for
observations missing race information.
53
Table 24. First year CFT tobit regression estimates by gender, FY05-FY12 accessions
Men
Women
Mean Coeff Derivative
Mean Coeff Derivative
(z)
(z)
IST score of 300
0.004 12.246** 12.2462 0.006 14.085** 14.0853
[7.20] [3.01]
IST middle third
0.335 -6.719** -6.7189 0.326 -6.516** -6.5157
[28.31] [7.84]
IST bottom third
0.334 -12.500** -12.5000 0.341 -13.015** -13.0151
[52.09] [15.76]
High quality
0.675 2.210** 2.2102 0.643 1.428* 1.4276
[10.59] [2.01]
DEP months >=3
0.780 0.369 0.3686 0.649 -0.866 -0.8660
[1.47] [1.14]
No enlistment waivers
0.554 0.008 0.0083 0.555 -0.725 -0.7251
[0.04] [1.06]
Age at accession
19.340 0.395** 0.3947 19.317 0.108 0.1076
[7.44] [0.62]
Enlistment bonus
0.175 0.396 0.3957 0.194 1.432 1.4323
[1.51] [1.60]
Accession trimester
Oct-Jan accession
0.305 -7.604** -7.6041 0.335 -6.108** -6.1075
54
Men
Women
Mean Coeff Derivative
Mean Coeff Derivative
(z)
(z)
[31.61] [7.16]
Feb-Apr accession
0.199 -4.342** -4.3422 0.254 -3.584** -3.5835
[16.12] [3.98]
Racial/ethnic
background
Asian
0.036 -0.663 -0.6628 0.042 0.588 0.5881
[1.29] [0.35]
Black
0.096 1.727** 1.7266 0.161 1.575+ 1.5748
[1.27] [0.78]
Hispanic
0.188 0.795** 0.7947 0.264 1.414+ 1.4143
[5.17] [1.67]
Parris Island
0.487 -2.597** -2.5970
[13.14]
Constant
293.708** 291.858**
[247.08] [76.15]
Number of observations
121,128
9,432
Chi square (df = 20 & 19)
21,837.3
1656.84
Absolute value of z statistics in brackets: + significant at 10%; * significant at 5%; **
significant at 1%. Note: These regressions also controlled for accession FY and for
observations missing race information.
55
Table 25. Second year CFT tobit regression estimates by gender, FY05-FY11
accessions
Men
Women
Mean Coeff Derivative
Mean Coeff Derivative
(z)
(z)
IST score of 300
0.003 13.162** 13.1617 0.005 15.863* 15.8631
[6.45] [2.50]
IST middle third
0.330 -6.976** -6.9762 0.332 -7.717** -7.7174
[25.21] [7.34]
IST bottom third
0.340 -14.537** -14.5365 0.338 -15.893** -15.8933
[52.06] [15.08]
High quality
0.650 1.908** 1.9083 0.630 1.501+ 1.5006
[8.01] [1.67]
DEP months >=3
0.714 -0.285 -0.2846 0.614 -0.422 -0.4220
[1.02] [0.44]
No enlistment waivers
0.529 0.758** 0.7576 0.550 -0.452 -0.4524
[3.31] [0.52]
Age at accession
19.385 0.301** 0.3008 19.312 -0.119 -0.1192
[5.00] [0.55]
Enlistment bonus
0.253 0.305 0.3047 0.281 1.344 1.3442
[1.13] [1.32]
56
Men
Women
Mean Coeff Derivative
Mean Coeff Derivative
(z)
(z)
Accession trimester
Oct-Jan accession
0.306 -6.499** -6.4985 0.328 -7.165** -7.1649
[23.09] [6.59]
Feb-Apr accession
0.197 -0.14 -0.1402 0.242 -0.957 -0.9570
[0.45] [0.83]
Racial/ethnic
background
Asian
0.035 -0.003 -0.0035 0.043 4.306* 4.3064
[0.01] [2.04]
Black
0.093 1.765** 1.7651 0.164 2.758* 2.7578
[1.23] [1.12]
Hispanic
0.182 1.809** 1.8093 0.260 4.158** 4.1579
[4.49] [2.32]
Parris Island
0.481 -3.565** -3.5646 N/A N/A N/A
[15.60] N/A
Constant
290.456** 291.226**
[19.85] [10.70]
Number of observations
115,292
8,319
Chi square (df = 21 & 19)
20,233.6
1538.89
Absolute value of z statistics in brackets: + significant at 10%; * significant at 5%; **
significant at 1%. Note: These regressions also controlled for accession FY and for
observations missing race information.
57
Table 26. DEP attrition logit regression estimates by gender, FY05-FY13 contractsa
Men Women
Mean Coeff Derivative Mean Coeff Derivative
(z) (z)
IST middle third
0.334 0.331** 0.0252
0.333 0.266** 0.0270
[19.57]
[5.48]
IST bottom third
0.339 0.862** 0.0788
0.335 0.903** 0.1093
[53.32]
[19.78]
High quality
0.645 0.197** 0.0175
0.618 0.275** 0.0318
[14.64]
[7.18]
No enlistment waivers
0.240 -0.005 -0.0005
0.226 -0.131** -0.0150
[0.33]
[2.88]
Enlistment bonus
0.159 -7.252** -0.1326
0.179 -7.094** -0.1960
[16.21]
[7.09]
Racial/ethnic
background
Asian
0.034 -0.112** -0.0098
0.039 -0.280** -0.0308
[3.17]
[2.78]
Black
0.092 -0.010 -0.0009
0.155 -0.015 -0.0017
[0.46]
[0.29]
Hispanic
0.180 -0.002 -0.0002
0.238 -0.228** -0.0260
58
Men Women
Mean Coeff Derivative Mean Coeff Derivative
(z) (z)
[0.10]
[5.14]
Constant
3.752**
4.715**
[42.36]
[17.99]
Number of observations
281,923
25,905
Chi square (df = 20 & 20)
28,248.3
3,967.6
Absolute value of z statistics in brackets: + significant at 10%; * significant at 5%; **
significant at 1%. Note: These regressions also controlled for accession FY and for
observations missing race information.
a. From Table 26 we see that men in the middle third of the IST score have a DEP attrition
rate that is 2.5 percentage points higher than similar men in the top third of the IST score;
and men in the bottom third of the IST score have a DEP attrition rate that is 7.9
percentage points higher than similar men in the top third of the IST score. For women,
these respective differences in attrition rates are 2.7 percentage points higher for women
in the middle third of the IST score relative to similar women in the top third of the IST score;
and 10.9 percentage points higher for women in the bottom third of the IST score relative
to similar women in the top third of the IST score. We had to exclude 48 percent of the DEP
attrites in this analysis since they did not have an IST score. This may bias the estimates
shown in Table 26.
The CNA Corporation
This report was written by CNA Corporation’s Resource Analysis Division
(RAD).
RAD provides analytical services—through empirical research, modeling,
and simulation—to help develop, evaluate, and implement policies,
practices, and programs that make people, budgets, and assets more
effective and efficient. Major areas of research include energy and
environment; manpower management; acquisition and cost;
infrastructure; health research policy; and military readiness.
DRM-2014-U-007869-Final
www.cna.org ● 703-824-2000
3003 Washington Boulevard, Arlington, VA 22201
CNA Corporation is a not-for-profit research organization
that serves the public interest by providing
in-depth analysis and result-oriented solutions
to help government leaders choose
the best course of action
in setting policy and managing operations.
Nobody gets closer—
to the people, to the data, to the problem.
top related